Compare commits

..

146 Commits

Author SHA1 Message Date
github-actions[bot]
4688b815b1 Proxy release 2025-05-14 09:39 UTC 2025-05-14 09:39:15 +00:00
github-actions[bot]
0982ca4636 Proxy release 2025-05-12 14:35 UTC 2025-05-12 14:35:27 +00:00
github-actions[bot]
7272d9f7b3 Proxy release 2025-05-06 09:47 UTC 2025-05-06 09:47:48 +00:00
github-actions[bot]
37d555aa59 Proxy release 2025/04/29 06:01 UTC 2025-04-29 06:01:28 +00:00
github-actions[bot]
cae3e2976b Proxy release 2025-04-22 2025-04-22 06:02:06 +00:00
github-actions[bot]
51ecd1bb37 Proxy release 2025-04-15 2025-04-15 06:01:10 +00:00
github-actions[bot]
1e6bb48076 Proxy release 2025-04-08 2025-04-08 06:01:37 +00:00
github-actions[bot]
1470af0b42 Proxy release 2025-04-01 2025-04-01 06:01:27 +00:00
github-actions[bot]
f92f92b91b Proxy release 2025-03-13 2025-03-13 13:43:01 +00:00
github-actions[bot]
dbb205ae92 Proxy release 2025-03-13 2025-03-13 09:50:35 +00:00
Conrad Ludgate
85072b715f Merge pull request #11106 from neondatabase/rc/release-proxy/2025-03-06
Proxy release 2025-03-06
2025-03-06 09:53:00 +00:00
github-actions[bot]
6c86fe7143 Proxy release 2025-03-06 2025-03-06 06:02:15 +00:00
JC Grünhage
66d5fe7f5b Merge pull request #11023 from neondatabase/rc/release-proxy/2025-02-27
Proxy release 2025-02-27
2025-02-27 19:10:58 +01:00
github-actions[bot]
a1b9528757 Proxy release 2025-02-27 2025-02-27 16:18:42 +00:00
Ivan Efremov
1423bb8aa2 Merge pull request #11011 from neondatabase/rc/release-proxy/2025-02-27
Proxy release 2025-02-27
2025-02-27 13:57:49 +02:00
github-actions[bot]
332f064a42 Proxy release 2025-02-27 2025-02-27 00:17:57 +00:00
Folke Behrens
c962f2b447 Merge pull request #10903 from neondatabase/rc/release-proxy/2025-02-20
Proxy release 2025-02-20
2025-02-20 10:37:47 +01:00
github-actions[bot]
446b3f9d28 Proxy release 2025-02-20 2025-02-20 06:02:01 +00:00
Conrad Ludgate
23352dc2e9 Merge pull request #10802 from neondatabase/rc/release-proxy/2025-02-13
Proxy release 2025-02-13
2025-02-13 08:41:01 +00:00
github-actions[bot]
c65fc5a955 Proxy release 2025-02-13 2025-02-13 06:02:01 +00:00
Ivan Efremov
3e624581cd Merge pull request #10691 from neondatabase/rc/release-proxy/2025-02-06
Proxy release 2025-02-06
2025-02-06 10:23:43 +02:00
github-actions[bot]
fedf4f169c Proxy release 2025-02-06 2025-02-06 06:02:11 +00:00
Folke Behrens
86d5798108 Merge pull request #10576 from neondatabase/rc/release-proxy/2025-01-30
Proxy release 2025-01-30
2025-01-30 08:52:09 +01:00
github-actions[bot]
8b4088dd8a Proxy release 2025-01-30 2025-01-30 06:02:00 +00:00
Ivan Efremov
c91905e643 Merge pull request #10416 from neondatabase/rc/release-proxy/2025-01-16
Proxy release 2025-01-16
2025-01-16 10:04:38 +02:00
github-actions[bot]
44b4e355a2 Proxy release 2025-01-16 2025-01-16 06:02:04 +00:00
Folke Behrens
03666a1f37 Merge pull request #10320 from neondatabase/rc/release-proxy/2025-01-09
Proxy release 2025-01-09
2025-01-09 10:19:07 +01:00
github-actions[bot]
9c92242ca0 Proxy release 2025-01-09 2025-01-09 06:02:06 +00:00
Conrad Ludgate
a354071dd0 Merge pull request #10180 from neondatabase/rc/release-proxy/2024-12-17
Proxy release 2024-12-17
2024-12-18 06:31:05 +00:00
github-actions[bot]
758680d4f8 Proxy release 2024-12-17 2024-12-17 22:06:42 +00:00
Conrad Ludgate
1738fd0a96 Merge pull request #10107 from neondatabase/rc/release-proxy/2024-12-12
Proxy release 2024-12-12
2024-12-12 10:21:30 +00:00
Conrad Ludgate
87b7edfc72 Merge branch 'release-proxy' into rc/release-proxy/2024-12-12 2024-12-12 09:58:31 +00:00
github-actions[bot]
def05700d5 Proxy release 2024-12-12 2024-12-12 06:02:08 +00:00
Ivan Efremov
b547681e08 Merge pull request #10024 from neondatabase/rc/release-proxy/2024-12-05
Proxy release 2024-12-05
2024-12-05 15:35:35 +02:00
Ivan Efremov
0fd211537b proxy: Present new auth backend cplane_proxy_v1 (#10012)
Implement a new auth backend based on the current Neon backend to switch
to the new Proxy V1 cplane API.

Implements [#21048](https://github.com/neondatabase/cloud/issues/21048)
2024-12-05 13:00:40 +02:00
Yuchen Liang
a83bd4e81c pageserver: fix buffered-writer on macos build (#10019)
## Problem

In https://github.com/neondatabase/neon/pull/9693, we forgot to check
macos build. The [CI
run](https://github.com/neondatabase/neon/actions/runs/12164541897/job/33926455468)
on main showed that macos build failed with unused variables and dead
code.

## Summary of changes

- add `allow(dead_code)` and `allow(unused_variables)` to the relevant
code that is not used on macos.

Signed-off-by: Yuchen Liang <yuchen@neon.tech>
2024-12-05 13:00:40 +02:00
Conrad Ludgate
ecdad5e6d5 chore: update rust-postgres (#10002)
Like #9931 but without rebasing upstream just yet, to try and minimise
the differences.

Removes all proxy-specific commits from the rust-postgres fork, now that
proxy no longer depends on them. Merging upstream changes to come later.
2024-12-05 13:00:40 +02:00
Conrad Ludgate
d028929945 chore: update clap (#10009)
This updates clap to use a new version of anstream
2024-12-05 13:00:40 +02:00
Yuchen Liang
7b0e3db868 pageserver: make BufferedWriter do double-buffering (#9693)
Closes #9387.

## Problem

`BufferedWriter` cannot proceed while the owned buffer is flushing to
disk. We want to implement double buffering so that the flush can happen
in the background. See #9387.

## Summary of changes

- Maintain two owned buffers in `BufferedWriter`.
- The writer is in charge of copying the data into owned, aligned
buffer, once full, submit it to the flush task.
- The flush background task is in charge of flushing the owned buffer to
disk, and returned the buffer to the writer for reuse.
- The writer and the flush background task communicate through a
bi-directional channel.

For in-memory layer, we also need to be able to read from the buffered
writer in `get_values_reconstruct_data`. To handle this case, we did the
following
- Use replace `VirtualFile::write_all` with `VirtualFile::write_all_at`,
and use `Arc` to share it between writer and background task.
- leverage `IoBufferMut::freeze` to get a cheaply clonable `IoBuffer`,
one clone will be submitted to the channel, the other clone will be
saved within the writer to serve reads. When we want to reuse the
buffer, we can invoke `IoBuffer::into_mut`, which gives us back the
mutable aligned buffer.
- InMemoryLayer reads is now aware of the maybe_flushed part of the
buffer.

**Caveat**

- We removed the owned version of write, because this interface does not
work well with buffer alignment. The result is that without direct IO
enabled,
[`download_object`](a439d57050/pageserver/src/tenant/remote_timeline_client/download.rs (L243))
does one more memcpy than before this PR due to the switch to use
`_borrowed` version of the write.
- "Bypass aligned part of write" could be implemented later to avoid
large amount of memcpy.

**Testing**
- use an oneshot channel based control mechanism to make flush behavior
deterministic in test.
- test reading from `EphemeralFile` when the last submitted buffer is
not flushed, in-progress, and done flushing to disk.


## Performance


We see performance improvement for small values, and regression on big
values, likely due to being CPU bound + disk write latency.


[Results](https://www.notion.so/neondatabase/Benchmarking-New-BufferedWriter-11-20-2024-143f189e0047805ba99acda89f984d51?pvs=4)


## Checklist before requesting a review

- [ ] I have performed a self-review of my code.
- [ ] If it is a core feature, I have added thorough tests.
- [ ] Do we need to implement analytics? if so did you add the relevant
metrics to the dashboard?
- [ ] If this PR requires public announcement, mark it with
/release-notes label and add several sentences in this section.

## Checklist before merging

- [ ] Do not forget to reformat commit message to not include the above
checklist

---------

Signed-off-by: Yuchen Liang <yuchen@neon.tech>
Co-authored-by: Christian Schwarz <christian@neon.tech>
2024-12-05 13:00:40 +02:00
John Spray
088eb72dd7 tests: make storcon scale test AZ-aware (#9952)
## Problem

We have a scale test for the storage controller which also acts as a
good stress test for scheduling stability. However, it created nodes
with no AZs set.

## Summary of changes

- Bump node count to 6 and set AZs on them.

This is a precursor to other AZ-related PRs, to make sure any new code
that's landed is getting scale tested in an AZ-aware environment.
2024-12-05 13:00:40 +02:00
a-masterov
d550e3f626 Create a branch for compute release (#9637)
## Problem
We practice a manual release flow for the compute module. This will
allow automation of the compute release process.

## Summary of changes
The workflow was modified to make a compute release automatically on the
branch release-compute.
## Checklist before requesting a review

- [x] I have performed a self-review of my code.
- [ ] If it is a core feature, I have added thorough tests.
- [ ] Do we need to implement analytics? if so did you add the relevant
metrics to the dashboard?
- [ ] If this PR requires public announcement, mark it with
/release-notes label and add several sentences in this section.

## Checklist before merging

- [ ] Do not forget to reformat commit message to not include the above
checklist
2024-12-05 13:00:40 +02:00
Erik Grinaker
8c6b41daf5 Display reqwest error source (#10004)
## Problem

Reqwest errors don't include details about the inner source error. This
means that we get opaque errors like:

```
receive body: error sending request for url (http://localhost:9898/v1/location_config)
```

Instead of the more helpful:

```
receive body: error sending request for url (http://localhost:9898/v1/location_config): operation timed out
```

Touches #9801.

## Summary of changes

Include the source error for `reqwest::Error` wherever it's displayed.
2024-12-05 13:00:40 +02:00
Alexey Kondratov
bbb050459b feat(compute): Set default application_name for pgbouncer connections (#9973)
## Problem

When client specifies `application_name`, pgbouncer propagates it to the
Postgres. Yet, if client doesn't do it, we have hard time figuring out
who opens a lot of Postgres connections (including the `cloud_admin`
ones).

See this investigation as an example:
https://neondb.slack.com/archives/C0836R0RZ0D

## Summary of changes

I haven't found this documented, but it looks like pgbouncer accepts
standard Postgres connstring parameters in the connstring in the
`[databases]` section, so put the default `application_name=pgbouncer`
there. That way, we will always see who opens Postgres connections. I
did tests, and if client specifies a `application_name`, pgbouncer
overrides this default, so it only works if it's not specified or set to
blank `&application_name=` in the connection string.

This is the last place we could potentially open some Postgres
connections without `application_name`. Everything else should be either
of two:
1. Direct client connections without `application_name`, but these
should be strictly non-`cloud_admin` ones
2. Some ad-hoc internal connections, so if we see spikes of unidentified
`cloud_admin` connections, we will need to investigate it again.

Fixes neondatabase/cloud#20948
2024-12-05 13:00:40 +02:00
Conrad Ludgate
cab498c787 feat(proxy): add option to forward startup params (#9979)
(stacked on #9990 and #9995)

Partially fixes #1287 with a custom option field to enable the fixed
behaviour. This allows us to gradually roll out the fix without silently
changing the observed behaviour for our customers.

related to https://github.com/neondatabase/cloud/issues/15284
2024-12-05 13:00:40 +02:00
Folke Behrens
6359342ffb Assign /libs/proxy/ to proxy team (#10003) 2024-12-05 13:00:40 +02:00
Erik Grinaker
13285c2a5e pageserver: return proper status code for heatmap_upload errors (#9991)
## Problem

During deploys, we see a lot of 500 errors due to heapmap uploads for
inactive tenants. These should be 503s instead.

Resolves #9574.

## Summary of changes

Make the secondary tenant scheduler use `ApiError` rather than
`anyhow::Error`, to propagate the tenant error and convert it to an
appropriate status code.
2024-12-05 13:00:40 +02:00
Peter Bendel
33790d14a3 fix parsing human time output like "50m37s" (#10001)
## Problem

In ingest_benchmark.yml workflow we use pgcopydb tool to migrate
project.
pgcopydb logs human time.

Our parsing of the human time doesn't work for times like "50m37s".

[Example
workflow](https://github.com/neondatabase/neon/actions/runs/12145539948/job/33867418065#step:10:479)

contains "57m45s"

but we
[reported](https://github.com/neondatabase/neon/actions/runs/12145539948/job/33867418065#step:10:500)
only the seconds part: 
45.000 s


## Summary of changes

add a regex pattern for Minute/Second combination
2024-12-05 13:00:40 +02:00
Peter Bendel
709b8cd371 optimize parms for ingest bench (#9999)
## Problem

we tried different parallelism settings for ingest bench 

## Summary of changes

the following settings seem optimal after merging
- SK side Wal filtering
- batched getpages

Settings:
- effective_io_concurrency 100
- concurrency limit 200 (different from Prod!)
- jobs 4, maintenance workers 7
- 10 GB chunk size
2024-12-05 13:00:40 +02:00
Vlad Lazar
1c9bbf1a92 storcon: return an error for drain attempts while paused (#9997)
## Problem

We currently allow drain operations to proceed while the node policy is
paused.

## Summary of changes

Return a precondition failed error in such cases. The orchestrator is
updated in https://github.com/neondatabase/infra/pull/2544 to skip drain
and fills if the pageserver is paused.

Closes: https://github.com/neondatabase/neon/issues/9907
2024-12-05 13:00:40 +02:00
Christian Schwarz
16163fb850 page_service: enable batching in Rust & Python Tests + Python benchmarks (#9993)
This is the first step towards batching rollout.

Refs

- rollout plan: https://github.com/neondatabase/cloud/issues/20620
- task https://github.com/neondatabase/neon/issues/9377
- uber-epic: https://github.com/neondatabase/neon/issues/9376
2024-12-05 13:00:40 +02:00
Alexander Bayandin
73ccc2b08c test_page_service_batching: fix non-numeric metrics (#9998)
## Problem

```
2024-12-03T15:42:46.5978335Z + poetry run python /__w/neon/neon/scripts/ingest_perf_test_result.py --ingest /__w/neon/neon/test_runner/perf-report-local
2024-12-03T15:42:49.5325077Z Traceback (most recent call last):
2024-12-03T15:42:49.5325603Z   File "/__w/neon/neon/scripts/ingest_perf_test_result.py", line 165, in <module>
2024-12-03T15:42:49.5326029Z     main()
2024-12-03T15:42:49.5326316Z   File "/__w/neon/neon/scripts/ingest_perf_test_result.py", line 155, in main
2024-12-03T15:42:49.5326739Z     ingested = ingest_perf_test_result(cur, item, recorded_at_timestamp)
2024-12-03T15:42:49.5327488Z                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2024-12-03T15:42:49.5327914Z   File "/__w/neon/neon/scripts/ingest_perf_test_result.py", line 99, in ingest_perf_test_result
2024-12-03T15:42:49.5328321Z     psycopg2.extras.execute_values(
2024-12-03T15:42:49.5328940Z   File "/github/home/.cache/pypoetry/virtualenvs/non-package-mode-_pxWMzVK-py3.11/lib/python3.11/site-packages/psycopg2/extras.py", line 1299, in execute_values
2024-12-03T15:42:49.5335618Z     cur.execute(b''.join(parts))
2024-12-03T15:42:49.5335967Z psycopg2.errors.InvalidTextRepresentation: invalid input syntax for type numeric: "concurrent-futures"
2024-12-03T15:42:49.5336287Z LINE 57:             'concurrent-futures',
2024-12-03T15:42:49.5336462Z                      ^
```

## Summary of changes
- `test_page_service_batching`: save non-numeric params as `labels`
- Add a runtime check that `metric_value` is NUMERIC
2024-12-05 13:00:40 +02:00
Christian Schwarz
c719be6474 tests & benchmarks: unify the way we customize the default tenant config (#9992)
Before this PR, some override callbacks used `.default()`, others
used `.setdefault()`.

As of this PR, all callbacks use `.setdefault()` which I think is least
prone to failure.

Aligning on a single way will set the right example for future tests
that need such customization.

The `test_pageserver_getpage_throttle.py` technically is a change in
behavior: before, it replaced the `tenant_config` field, now it just
configures the throttle. This is what I believe is intended anyway.
2024-12-05 13:00:40 +02:00
Arpad Müller
718645e56c Support tenant manifests in the scrubber (#9942)
Support tenant manifests in the storage scrubber:

* list the manifests, order them by generation
* delete all manifests except for the two most recent generations
* for the latest manifest: try parsing it.

I've tested this patch by running the against a staging bucket and it
successfully deleted stuff (and avoided deleting the latest two
generations).

In follow-up work, we might want to also check some invariants of the
manifest, as mentioned in #8088.

Part of #9386
Part of #8088

---------

Co-authored-by: Christian Schwarz <christian@neon.tech>
2024-12-05 13:00:40 +02:00
Conrad Ludgate
fbc8c36983 chore(proxy): enforce single host+port (#9995)
proxy doesn't ever provide multiple hosts/ports, so this code adds a lot
of complexity of error handling for no good reason.

(stacked on #9990)
2024-12-05 13:00:40 +02:00
Alexey Immoreev
5519e42612 Improvement: add console redirect timeout warning (#9985)
## Problem

There is no information on session being cancelled in 2 minutes at the
moment

## Summary of changes

The timeout being logged for the user
2024-12-05 13:00:40 +02:00
Erik Grinaker
4157eaf4c5 pageserver: respond to multiple shutdown signals (#9982)
## Problem

The Pageserver signal handler would only respond to a single signal and
initiate shutdown. Subsequent signals were ignored. This meant that a
`SIGQUIT` sent after a `SIGTERM` had no effect (e.g. in the case of a
slow or stalled shutdown). The `test_runner` uses this to force shutdown
if graceful shutdown is slow.

Touches #9740.

## Summary of changes

Keep responding to signals after the initial shutdown signal has been
received.

Arguably, the `test_runner` should also use `SIGKILL` rather than
`SIGQUIT` in this case, but it seems reasonable to respond to `SIGQUIT`
regardless.
2024-12-05 13:00:40 +02:00
Conrad Ludgate
60241127e2 chore(proxy): remove postgres config parser and md5 support (#9990)
Keeping the `mock` postgres cplane adaptor using "stock" tokio-postgres
allows us to remove a lot of dead weight from our actual postgres
connection logic.
2024-12-05 13:00:40 +02:00
John Spray
f7d5322e8b pageserver: more detailed logs when calling re-attach (#9996)
## Problem

We saw a peculiar case where a pageserver apparently got a 0-tenant
response to `/re-attach` but we couldn't see the request landing on a
storage controller. It was hard to confirm retrospectively that the
pageserver was configured properly at the moment it sent the request.

## Summary of changes

- Log the URL to which we are sending the request
- Log the NodeId and metadata that we sent
2024-12-05 13:00:40 +02:00
John Spray
41bb9c5280 pageserver: only store SLRUs & aux files on shard zero (#9786)
## Problem

Since https://github.com/neondatabase/neon/pull/9423 the non-zero shards
no longer need SLRU content in order to do GC. This data is now
redundant on shards >0.

One release cycle after merging that PR, we may merge this one, which
also stops writing those pages to shards > 0, reaping the efficiency
benefit.

Closes: https://github.com/neondatabase/neon/issues/7512
Closes: https://github.com/neondatabase/neon/issues/9641

## Summary of changes

- Avoid storing SLRUs on non-zero shards
- Bonus: avoid storing aux files on non-zero shards
2024-12-05 13:00:40 +02:00
John Spray
69c0d61c5c storcon: in shard splits, inherit parent's AZ (#9946)
## Problem

Sharded tenants should be run in a single AZ for best performance, so
that computes have AZ-local latency to all the shards.

Part of https://github.com/neondatabase/neon/issues/8264

## Summary of changes

- When we split a tenant, instead of updating each shard's preferred AZ
to wherever it is scheduled, propagate the preferred AZ from the parent.
- Drop the check in `test_shard_preferred_azs` that asserts shards end
up in their preferred AZ: this will not be true again until the
optimize_attachment logic is updated to make this so. The existing check
wasn't testing anything about scheduling, it was just asserting that we
set preferred AZ in a way that matches the way things happen to be
scheduled at time of split.
2024-12-05 13:00:40 +02:00
Christian Schwarz
63cb8ce975 pageserver: only throttle pagestream requests & bring back throttling deduction for smgr latency metrics (#9962)
## Problem

In the batching PR 
- https://github.com/neondatabase/neon/pull/9870

I stopped deducting the time-spent-in-throttle fro latency metrics,
i.e.,
- smgr latency metrics (`SmgrOpTimer`)
- basebackup latency (+scan latency, which I think is part of
basebackup).

The reason for stopping the deduction was that with the introduction of
batching, the trick with tracking time-spent-in-throttle inside
RequestContext and swap-replacing it from the `impl Drop for
SmgrOpTimer` no longer worked with >1 requests in a batch.

However, deducting time-spent-in-throttle is desirable because our
internal latency SLO definition does not account for throttling.

## Summary of changes

- Redefine throttling to be a page_service pagestream request throttle
instead of a throttle for repository `Key` reads through `Timeline::get`
/ `Timeline::get_vectored`.
- This means reads done by `basebackup` are no longer subject to any
throttle.
- The throttle applies after batching, before handling of the request.
- Drive-by fix: make throttle sensitive to cancellation.
- Rename metric label `kind` from `timeline_get` to `pagestream` to
reflect the new scope of throttling.

To avoid config format breakage, we leave the config field named
`timeline_get_throttle` and ignore the `task_kinds` field.
This will be cleaned up in a future PR.

## Trade-Offs

Ideally, we would apply the throttle before reading a request off the
connection, so that we queue the minimal amount of work inside the
process.
However, that's not possible because we need to do shard routing.

The redefinition of the throttle to limit pagestream request rate
instead of repository `Key` rate comes with several downsides:
- We're no longer able to use the throttle mechanism for other other
tasks, e.g. image layer creation.
  However, in practice, we never used that capability anyways.
- We no longer throttle basebackup.
2024-12-05 13:00:40 +02:00
Erik Grinaker
907e4aa3c4 test_runner: use immediate shutdown in test_sharded_ingest (#9984)
## Problem

`test_sharded_ingest` ingests a lot of data, which can cause shutdown to
be slow e.g. due to local "S3 uploads" or compactions. This can cause
test flakes during teardown.

Resolves #9740.

## Summary of changes

Perform an immediate shutdown of the cluster.
2024-12-05 13:00:40 +02:00
Erik Grinaker
0a2a84b766 safekeeper,pageserver: add heap profiling (#9778)
## Problem

We don't have good observability for memory usage. This would be useful
e.g. to debug OOM incidents or optimize performance or resource usage.

We would also like to use continuous profiling with e.g. [Grafana Cloud
Profiles](https://grafana.com/products/cloud/profiles-for-continuous-profiling/)
(see https://github.com/neondatabase/cloud/issues/14888).

This PR is intended as a proof of concept, to try it out in staging and
drive further discussions about profiling more broadly.

Touches https://github.com/neondatabase/neon/issues/9534.
Touches https://github.com/neondatabase/cloud/issues/14888.
Depends on #9779.
Depends on #9780.

## Summary of changes

Adds a HTTP route `/profile/heap` that takes a heap profile and returns
it. Query parameters:

* `format`: output format (`jemalloc` or `pprof`; default `pprof`).

Unlike CPU profiles (see #9764), heap profiles are not symbolized and
require the original binary to translate addresses to function names. To
make this work with Grafana, we'll probably have to symbolize the
process server-side -- this is left as future work, as is other output
formats like SVG.

Heap profiles don't work on macOS due to limitations in jemalloc.
2024-12-05 13:00:40 +02:00
a-masterov
85b12ddd52 Add support for the extensions test for Postgres v17 (#9748)
## Problem
The extensions for Postgres v17 are ready but we do not test the
extensions shipped with v17
## Summary of changes
Build the test image based on Postgres v17. Run the tests for v17.

---------

Co-authored-by: Anastasia Lubennikova <anastasia@neon.tech>
2024-12-05 13:00:40 +02:00
Christian Schwarz
dd76f1eeee page_service: batching observability & include throttled time in smgr metrics (#9870)
This PR 

- fixes smgr metrics https://github.com/neondatabase/neon/issues/9925 
- adds an additional startup log line logging the current batching
config
- adds a histogram of batch sizes global and per-tenant
- adds a metric exposing the current batching config

The issue described #9925 is that before this PR, request latency was
only observed *after* batching.
This means that smgr latency metrics (most importantly getpage latency)
don't account for
- `wait_lsn` time 
- time spent waiting for batch to fill up / the executor stage to pick
up the batch.

The fix is to use a per-request batching timer, like we did before the
initial batching PR.
We funnel those timers through the entire request lifecycle.

I noticed that even before the initial batching changes, we weren't
accounting for the time spent writing & flushing the response to the
wire.
This PR drive-by fixes that deficiency by dropping the timers at the
very end of processing the batch, i.e., after the `pgb.flush()` call.

I was **unable to maintain the behavior that we deduct
time-spent-in-throttle from various latency metrics.
The reason is that we're using a *single* counter in `RequestContext` to
track micros spent in throttle.
But there are *N* metrics timers in the batch, one per request.
As a consequence, the practice of consuming the counter in the drop
handler of each timer no longer works because all but the first timer
will encounter error `close() called on closed state`.
A failed attempt to maintain the current behavior can be found in
https://github.com/neondatabase/neon/pull/9951.

So, this PR remvoes the deduction behavior from all metrics.
I started a discussion on Slack about it the implications this has for
our internal SLO calculation:
https://neondb.slack.com/archives/C033RQ5SPDH/p1732910861704029

# Refs

- fixes https://github.com/neondatabase/neon/issues/9925
- sub-issue https://github.com/neondatabase/neon/issues/9377
- epic: https://github.com/neondatabase/neon/issues/9376
2024-12-05 13:00:40 +02:00
Christian Schwarz
8963ac85f9 storcon_cli tenant-describe: include tenant-wide information in output (#9899)
Before this PR, the storcon_cli didn't have a way to show the
tenant-wide information of the TenantDescribeResponse.

Sadly, the `Serialize` impl for the tenant config doesn't skip on
`None`, so, the output becomes a bit bloated.
Maybe we can use `skip_serializing_if(Option::is_none)` in the future.
=> https://github.com/neondatabase/neon/issues/9983
2024-12-05 13:00:40 +02:00
John Spray
4a488b3e24 storcon: use proper schedule context during node delete (#9958)
## Problem

I was touching `test_storage_controller_node_deletion` because for AZ
scheduling work I was adding a change to the storage controller (kick
secondaries during optimisation) that made a FIXME in this test defunct.
While looking at it I also realized that we can easily fix the way node
deletion currently doesn't use a proper ScheduleContext, using the
iterator type recently added for that purpose.

## Summary of changes

- A testing-only behavior in storage controller where if a secondary
location isn't yet ready during optimisation, it will be actively
polled.
- Remove workaround in `test_storage_controller_node_deletion` that
previously was needed because optimisation would get stuck on cold
secondaries.
- Update node deletion code to use a `TenantShardContextIterator` and
thereby a proper ScheduleContext
2024-12-05 13:00:40 +02:00
Alexey Kondratov
c4987b0b13 fix(testing): Use 1 MB shared_buffers even with LFC (#9969)
## Problem

After enabling LFC in tests and lowering `shared_buffers` we started
having more problems with `test_pg_regress`.

## Summary of changes

Set `shared_buffers` to 1MB to both exercise getPage requests/LFC, and
still have enough room for Postgres to operate. Everything smaller might
be not enough for Postgres under load, and can cause errors like 'no
unpinned buffers available'.

See Konstantin's comment [1] as well.

Fixes #9956

[1]:
https://github.com/neondatabase/neon/issues/9956#issuecomment-2511608097
2024-12-05 13:00:40 +02:00
Tristan Partin
84b4821118 Stop changing the value of neon.extension_server_port at runtime (#9972)
On reconfigure, we no longer passed a port for the extension server
which caused us to not write out the neon.extension_server_port line.
Thus, Postgres thought we were setting the port to the default value of
0. PGC_POSTMASTER GUCs cannot be set at runtime, which causes the
following log messages:

> LOG: parameter "neon.extension_server_port" cannot be changed without
restarting the server
> LOG: configuration file
"/var/db/postgres/compute/pgdata/postgresql.conf" contains errors;
unaffected changes were applied

Fixes: https://github.com/neondatabase/neon/issues/9945

Signed-off-by: Tristan Partin <tristan@neon.tech>
2024-12-05 13:00:40 +02:00
Conrad Ludgate
32ba9811f9 feat(proxy): emit JWT auth method and JWT issuer in parquet logs (#9971)
Fix the HTTP AuthMethod to accomodate the JWT authorization method.
Introduces the JWT issuer as an additional field in the parquet logs
2024-12-05 13:00:40 +02:00
Folke Behrens
a0cd64c4d3 Bump OTel, tracing, reqwest crates (#9970) 2024-12-05 13:00:40 +02:00
Arseny Sher
84687b743d Update consensus protocol spec (#9607)
The spec was written for the buggy protocol which we had before the one
more similar to Raft was implemented. Update the spec with what we
currently have.

ref https://github.com/neondatabase/neon/issues/8699
2024-12-05 13:00:40 +02:00
Folke Behrens
b6f93dcec9 proxy: Create Elasticache credentials provider lazily (#9967)
## Problem

The credentials providers tries to connect to AWS STS even when we use
plain Redis connections.

## Summary of changes

* Construct the CredentialsProvider only when needed ("irsa").
2024-12-05 13:00:40 +02:00
Alexander Bayandin
4f6c594973 CI(replication-tests): fix notifications about replication-tests failures (#9950)
## Problem

`if: ${{ github.event.schedule }}` gets skipped if a previous step has
failed, but we want to run the step for both `success` and `failure`

## Summary of changes
- Add `!cancelled()` to notification step if-condition, to skip only
cancelled jobs
2024-12-05 13:00:40 +02:00
Conrad Ludgate
a750c14735 fix(proxy): forward notifications from authentication (#9948)
Fixes https://github.com/neondatabase/cloud/issues/20973. 

This refactors `connect_raw` in order to return direct access to the
delayed notices.

I cannot find a way to test this with psycopg2 unfortunately, although
testing it with psql does return the expected results.
2024-12-05 13:00:40 +02:00
John Spray
9ce0dd4e55 storcon: add metric for AZ scheduling violations (#9949)
## Problem

We can't easily tell how far the state of shards is from their AZ
preferences. This can be a cause of performance issues, so it's
important for diagnosability that we can tell easily if there are
significant numbers of shards that aren't running in their preferred AZ.

Related: https://github.com/neondatabase/cloud/issues/15413

## Summary of changes

- In reconcile_all, count shards that are scheduled into the wrong AZ
(if they have a preference), and publish it as a prometheus gauge.
- Also calculate a statistic for how many shards wanted to reconcile but
couldn't.

This is clearly a lazy calculation: reconcile all only runs
periodically. But that's okay: shards in the wrong AZ is something that
only matters if it stays that way for some period of time.
2024-12-05 13:00:40 +02:00
Erik Grinaker
0e1a336607 test_runner: improve wait_until (#9936)
Improves `wait_until` by:

* Use `timeout` instead of `iterations`. This allows changing the
timeout/interval parameters independently.
* Make `timeout` and `interval` optional (default 20s and 0.5s). Most
callers don't care.
* Only output status every 1s by default, and add optional
`status_interval` parameter.
* Remove `show_intermediate_error`, this was always emitted anyway.

Most callers have been updated to use the defaults, except where they
had good reason otherwise.
2024-12-05 13:00:40 +02:00
Anastasia Lubennikova
7fc2912d06 Update pgvector to 0.8.0 (#9733) 2024-12-05 13:00:40 +02:00
John Spray
fdf231c237 storcon: don't take any Service locks in /status and /ready (#9944)
## Problem

We saw unexpected container terminations when running in k8s with with
small CPU resource requests.

The /status and /ready handlers called `maybe_forward`, which always
takes the lock on Service::inner.

If there is a lot of writer lock contention, and the container is
starved of CPU, this increases the likelihood that we will get killed by
the kubelet.

It isn't certain that this was a cause of issues, but it is a potential
source that we can eliminate.

## Summary of changes

- Revise logic to return immediately if the URL is in the non-forwarded
list, rather than calling maybe_forward
2024-12-05 13:00:40 +02:00
Konstantin Knizhnik
1e08b5dccc Fix issues with prefetch ring buffer resize (#9847)
## Problem

See https://neondb.slack.com/archives/C04DGM6SMTM/p1732110190129479


We observe the following error in the logs 
```
[XX000] ERROR: [NEON_SMGR] [shard 3] Incorrect prefetch read: status=1 response=0x7fafef335138 my=128 receive=128
```
most likely caused by changing `neon.readahead_buffer_size`

## Summary of changes

1. Copy shard state
2. Do not use prefetch_set_unused in readahead_buffer_resize
3. Change prefetch buffer overflow criteria

---------

Co-authored-by: Konstantin Knizhnik <knizhnik@neon.tech>
2024-12-05 13:00:40 +02:00
Alexander Bayandin
030810ed3e Compute image: prepare Postgres v14-v16 for Debian 12 (#9954)
## Problem

Current compute images for Postgres 14-16 don't build on Debian 12
because of issues with extensions.
This PR fixes that, but for the current setup, it is mostly a no-op
change.

## Summary of changes
- Use `/bin/bash -euo pipefail` as SHELL to fail earlier
- Fix `plv8` build: backport a trivial patch for v8
- Fix `postgis` build: depend `sfgal` version on Debian version instead
of Postgres version


Tested in: https://github.com/neondatabase/neon/pull/9849
2024-12-05 13:00:40 +02:00
Konstantin Knizhnik
62b74bdc2c Add GUC controlling whether to pause recovery if some critical GUCs at replica have smaller value than on primary (#9057)
## Problem

See https://github.com/neondatabase/neon/issues/9023

## Summary of changes

Ass GUC `recovery_pause_on_misconfig` allowing not to pause in case of
replica and primary configuration mismatch

See https://github.com/neondatabase/postgres/pull/501
See https://github.com/neondatabase/postgres/pull/502
See https://github.com/neondatabase/postgres/pull/503
See https://github.com/neondatabase/postgres/pull/504


## Checklist before requesting a review

- [ ] I have performed a self-review of my code.
- [ ] If it is a core feature, I have added thorough tests.
- [ ] Do we need to implement analytics? if so did you add the relevant
metrics to the dashboard?
- [ ] If this PR requires public announcement, mark it with
/release-notes label and add several sentences in this section.

## Checklist before merging

- [ ] Do not forget to reformat commit message to not include the above
checklist

---------

Co-authored-by: Konstantin Knizhnik <knizhnik@neon.tech>
Co-authored-by: Heikki Linnakangas <heikki@neon.tech>
2024-12-05 13:00:40 +02:00
Folke Behrens
8b7e9ed820 Merge the consumption metric pushes (#9939)
#8564

## Problem

The main and backup consumption metric pushes are completely
independent,
resulting in different event time windows and different idempotency
keys.

## Summary of changes

* Merge the push tasks, but keep chunks the same size.
2024-12-05 13:00:40 +02:00
Christian Schwarz
5dad89acd4 page_service: rewrite batching to work without a timeout (#9851)
# Problem

The timeout-based batching adds latency to unbatchable workloads.

We can choose a short batching timeout (e.g. 10us) but that requires
high-resolution timers, which tokio doesn't have.
I thoroughly explored options to use OS timers (see
[this](https://github.com/neondatabase/neon/pull/9822) abandoned PR).
In short, it's not an attractive option because any timer implementation
adds non-trivial overheads.

# Solution

The insight is that, in the steady state of a batchable workload, the
time we spend in `get_vectored` will be hundreds of microseconds anyway.

If we prepare the next batch concurrently to `get_vectored`, we will
have a sizeable batch ready once `get_vectored` of the current batch is
done and do not need an explicit timeout.

This can be reasonably described as **pipelining of the protocol
handler**.

# Implementation

We model the sub-protocol handler for pagestream requests
(`handle_pagrequests`) as two futures that form a pipeline:

2. Batching: read requests from the connection and fill the current
batch
3. Execution: `take` the current batch, execute it using `get_vectored`,
and send the response.

The Reading and Batching stage are connected through a new type of
channel called `spsc_fold`.

See the long comment in the `handle_pagerequests_pipelined` for details.

# Changes

- Refactor `handle_pagerequests`
    - separate functions for
- reading one protocol message; produces a `BatchedFeMessage` with just
one page request in it
- batching; tried to merge an incoming `BatchedFeMessage` into an
existing `BatchedFeMessage`; returns `None` on success and returns back
the incoming message in case merging isn't possible
        - execution of a batched message
- unify the timeline handle acquisition & request span construction; it
now happen in the function that reads the protocol message
- Implement serial and pipelined model
    - serial: what we had before any of the batching changes
      - read one protocol message
      - execute protocol messages
    - pipelined: the design described above
- optionality for execution of the pipeline: either via concurrent
futures vs tokio tasks
- Pageserver config
  - remove batching timeout field
  - add ability to configure pipelining mode
- add ability to limit max batch size for pipelined configurations
(required for the rollout, cf
https://github.com/neondatabase/cloud/issues/20620 )
  - ability to configure execution mode
- Tests
  - remove `batch_timeout` parametrization
  - rename `test_getpage_merge_smoke` to `test_throughput`
- add parametrization to test different max batch sizes and execution
moes
  - rename `test_timer_precision` to `test_latency`
  - rename the test case file to `test_page_service_batching.py`
  - better descriptions of what the tests actually do

## On the holding The `TimelineHandle` in the pending batch

While batching, we hold the `TimelineHandle` in the pending batch.
Therefore, the timeline will not finish shutting down while we're
batching.

This is not a problem in practice because the concurrently ongoing
`get_vectored` call will fail quickly with an error indicating that the
timeline is shutting down.
This results in the Execution stage returning a `QueryError::Shutdown`,
which causes the pipeline / entire page service connection to shut down.
This drops all references to the
`Arc<Mutex<Option<Box<BatchedFeMessage>>>>` object, thereby dropping the
contained `TimelineHandle`s.

- => fixes https://github.com/neondatabase/neon/issues/9850

# Performance

Local run of the benchmarks, results in [this empty
commit](1cf5b1463f)
in the PR branch.

Key take-aways:
* `concurrent-futures` and `tasks` deliver identical `batching_factor`
* tail latency impact unknown, cf
https://github.com/neondatabase/neon/issues/9837
* `concurrent-futures` has higher throughput than `tasks` in all
workloads (=lower `time` metric)
* In unbatchable workloads, `concurrent-futures` has 5% higher
`CPU-per-throughput` than that of `tasks`, and 15% higher than that of
`serial`.
* In batchable-32 workload, `concurrent-futures` has 8% lower
`CPU-per-throughput` than that of `tasks` (comparison to tput of
`serial` is irrelevant)
* in unbatchable workloads, mean and tail latencies of
`concurrent-futures` is practically identical to `serial`, whereas
`tasks` adds 20-30us of overhead

Overall, `concurrent-futures` seems like a slightly more attractive
choice.

# Rollout

This change is disabled-by-default.

Rollout plan:
- https://github.com/neondatabase/cloud/issues/20620

# Refs

- epic: https://github.com/neondatabase/neon/issues/9376
- this sub-task: https://github.com/neondatabase/neon/issues/9377
- the abandoned attempt to improve batching timeout resolution:
https://github.com/neondatabase/neon/pull/9820
- closes https://github.com/neondatabase/neon/issues/9850
- fixes https://github.com/neondatabase/neon/issues/9835
2024-12-05 13:00:40 +02:00
Matthias van de Meent
547b2d2827 Fix timeout value used in XLogWaitForReplayOf (#9937)
The previous value assumed usec precision, while the timeout used is in
milliseconds, causing replica backends to wait for (potentially) many
hours for WAL replay without the expected progress reports in logs.

This fixes the issue.

Reported-By: Alexander Lakhin <exclusion@gmail.com>

## Problem


https://github.com/neondatabase/postgres/pull/279#issuecomment-2507671817

The timeout value was configured with the assumption the indicated value
would be microseconds, where it's actually milliseconds. That causes the
backend to wait for much longer (2h46m40s) before it emits the "I'm
waiting for recovery" message. While we do have wait events configured
on this, it's not great to have stuck backends without clear logs, so
this fixes the timeout value in all our PostgreSQL branches.

## PG PRs

* PG14: https://github.com/neondatabase/postgres/pull/542
* PG15: https://github.com/neondatabase/postgres/pull/543
* PG16: https://github.com/neondatabase/postgres/pull/544
* PG17: https://github.com/neondatabase/postgres/pull/545
2024-12-05 13:00:40 +02:00
Gleb Novikov
93f29a0065 Fixed fast_import pgbin in calling get_pg_version (#9933)
Was working on https://github.com/neondatabase/cloud/pull/20795 and
discovered that fast_import is not working normally.
2024-12-05 13:00:40 +02:00
John Spray
4f36494615 pageserver: download small objects using a smaller timeout (#9938)
## Problem

It appears that the Azure storage API tends to hang TCP connections more
than S3 does.

Currently we use a 2 minute timeout for all downloads. This is large
because sometimes the objects we download are large. However, waiting 2
minutes when doing something like downloading a manifest on tenant
attach is problematic, because when someone is doing a "create tenant,
create timeline" workflow, that 2 minutes is long enough for them
reasonably to give up creating that timeline.

Rather than propagate oversized timeouts further up the stack, we should
use a different timeout for objects that we expect to be small.

Closes: https://github.com/neondatabase/neon/issues/9836

## Summary of changes

- Add a `small_timeout` configuration attribute to remote storage,
defaulting to 30 seconds (still a very generous period to do something
like download an index)
- Add a DownloadKind parameter to DownloadOpts, so that callers can
indicate whether they expect the object to be small or large.
- In the azure client, use small timeout for HEAD requests, and for GET
requests if DownloadKind::Small is used.
- Use DownloadKind::Small for manifests, indices, and heatmap downloads.

This PR intentionally does not make the equivalent change to the S3
client, to reduce blast radius in case this has unexpected consequences
(we could accomplish the same thing by editing lots of configs, but just
skipping the code is simpler for right now)
2024-12-05 13:00:40 +02:00
Alexey Kondratov
0a550f3e7d feat(compute_ctl): Always set application_name (#9934)
## Problem

It was not always possible to judge what exactly some `cloud_admin`
connections were doing because we didn't consistently set
`application_name` everywhere.

## Summary of changes

Unify the way we connect to Postgres:
1. Switch to building configs everywhere
2. Always set `application_name` and make naming consistent

Follow-up for #9919
Part of neondatabase/cloud#20948
2024-12-05 13:00:40 +02:00
Erik Grinaker
4bb9554e4a safekeeper: use jemalloc (#9780)
## Problem

To add Safekeeper heap profiling in #9778, we need to switch to an
allocator that supports it. Pageserver and proxy already use jemalloc.

Touches #9534.

## Summary of changes

Use jemalloc in Safekeeper.
2024-12-05 13:00:40 +02:00
John Spray
008616cfe6 storage controller: use proper ScheduleContext when evacuating a node (#9908)
## Problem

When picking locations for a shard, we should use a ScheduleContext that
includes all the other shards in the tenant, so that we apply proper
anti-affinity between shards. If we don't do this, then it can lead to
unstable scheduling, where we place a shard somewhere that the optimizer
will then immediately move it away from.

We didn't always do this, because it was a bit awkward to accumulate the
context for a tenant rather than just walking tenants.

This was a TODO in `handle_node_availability_transition`:
```
                        // TODO: populate a ScheduleContext including all shards in the same tenant_id (only matters
                        // for tenants without secondary locations: if they have a secondary location, then this
                        // schedule() call is just promoting an existing secondary)
```

This is a precursor to https://github.com/neondatabase/neon/issues/8264,
where the current imperfect scheduling during node evacuation hampers
testing.

## Summary of changes

- Add an iterator type that yields each shard along with a
schedulecontext that includes all the other shards from the same tenant
- Use the iterator to replace hand-crafted logic in optimize_all_plan
(functionally identical)
- Use the iterator in `handle_node_availability_transition` to apply
proper anti-affinity during node evacuation.
2024-12-05 13:00:40 +02:00
Conrad Ludgate
e61ec94fbc chore(proxy): vendor a subset of rust-postgres (#9930)
Our rust-postgres fork is getting messy. Mostly because proxy wants more
control over the raw protocol than tokio-postgres provides. As such,
it's diverging more and more. Storage and compute also make use of
rust-postgres, but in more normal usage, thus they don't need our crazy
changes.

Idea: 
* proxy maintains their subset
* other teams use a minimal patch set against upstream rust-postgres

Reviewing this code will be difficult. To implement it, I
1. Copied tokio-postgres, postgres-protocol and postgres-types from
00940fcdb5
2. Updated their package names with the `2` suffix to make them compile
in the workspace.
3. Updated proxy to use those packages
4. Copied in the code from tokio-postgres-rustls 0.13 (with some patches
applied https://github.com/jbg/tokio-postgres-rustls/pull/32
https://github.com/jbg/tokio-postgres-rustls/pull/33)
5. Removed as much dead code as I could find in the vendored libraries
6. Updated the tokio-postgres-rustls code to use our existing channel
binding implementation
2024-12-05 13:00:40 +02:00
Erik Grinaker
e5152551ad test_runner/performance: add logical message ingest benchmark (#9749)
Adds a benchmark for logical message WAL ingestion throughput
end-to-end. Logical messages are essentially noops, and thus ignored by
the Pageserver.

Example results from my MacBook, with fsync enabled:

```
postgres_ingest: 14.445 s
safekeeper_ingest: 29.948 s
pageserver_ingest: 30.013 s
pageserver_recover_ingest: 8.633 s
wal_written: 10,340 MB
message_count: 1310720 messages
postgres_throughput: 715 MB/s
safekeeper_throughput: 345 MB/s
pageserver_throughput: 344 MB/s
pageserver_recover_throughput: 1197 MB/s
```

See
https://github.com/neondatabase/neon/issues/9642#issuecomment-2475995205
for running analysis.

Touches #9642.
2024-12-05 13:00:40 +02:00
Alexey Kondratov
b0822a5499 fix(compute_ctl): Allow usage of DB names with whitespaces (#9919)
## Problem

We used `set_path()` to replace the database name in the connection
string. It automatically does url-safe encoding if the path is not
already encoded, but it does it as per the URL standard, which assumes
that tabs can be safely removed from the path without changing the
meaning of the URL. See, e.g.,
https://url.spec.whatwg.org/#concept-basic-url-parser. It also breaks
for DBs with properly %-encoded names, like with `%20`, as they are kept
intact, but actually should be escaped.

Yet, this is not true for Postgres, where it's completely valid to have
trailing tabs in the database name.

I think this is the PR that caused this regression
https://github.com/neondatabase/neon/pull/9717, as it switched from
`postgres::config::Config` back to `set_path()`.

This was fixed a while ago already [1], btw, I just haven't added a test
to catch this regression back then :(

## Summary of changes

This commit changes the code back to use
`postgres/tokio_postgres::Config` everywhere.

While on it, also do some changes around, as I had to touch this code:
1. Bump some logging from `debug` to `info` in the spec apply path. We
do not use `debug` in prod, and it was tricky to understand what was
going on with this bug in prod.
2. Refactor configuration concurrency calculation code so it was
reusable. Yet, still keep `1` in the case of reconfiguration. The
database can be actively used at this moment, so we cannot guarantee
that there will be enough spare connection slots, and the underlying
code won't handle connection errors properly.
3. Simplify the installed extensions code. It was spawning a blocking
task inside async function, which doesn't make much sense. Instead, just
have a main sync function and call it with `spawn_blocking` in the API
code -- the only place we need it to be async.
4. Add regression python test to cover this and related problems in the
future. Also, add more extensive testing of schema dump and DBs and
roles listing API.

[1]:
4d1e48f3b9
[2]:
https://www.postgresql.org/message-id/flat/20151023003445.931.91267%40wrigleys.postgresql.org

Resolves neondatabase/cloud#20869
2024-12-05 13:00:40 +02:00
Alexander Bayandin
1fb6ab59e8 test_runner: rerun all failed tests (#9917)
## Problem

Currently, we rerun only known flaky tests. This approach was chosen to
reduce the number of tests that go unnoticed (by forcing people to take
a look at failed tests and rerun the job manually), but it has some
drawbacks:
- In PRs, people tend to push new changes without checking failed tests
(that's ok)
- In the main, tests are just restarted without checking
(understandable)
- Parametrised tests become flaky one by one, i.e. if `test[1]` is flaky
`, test[2]` is not marked as flaky automatically (which may or may not
be the case).

I suggest rerunning all failed tests to increase the stability of GitHub
jobs and using the Grafana Dashboard with flaky tests for deeper
analysis.

## Summary of changes
- Rerun all failed tests twice at max
2024-12-05 13:00:40 +02:00
Vlad Lazar
e16439400d pageserver: return correct LSN for interpreted proto keep alive responses (#9928)
## Problem

For the interpreted proto the pageserver is not returning the correct
LSN
in replies to keep alive requests. This is because the interpreted
protocol arm
was not updating `last_rec_lsn`.

## Summary of changes

* Return correct LSN in keep-alive responses
* Fix shard field in wal sender traces
2024-12-05 13:00:40 +02:00
Arpad Müller
e401f66698 Update rust to 1.83.0, also update cargo adjacent tools (#9926)
We keep the practice of keeping the compiler up to date, pointing to the
latest release. This is done by many other projects in the Rust
ecosystem as well.

[Release notes](https://releases.rs/docs/1.83.0/).

Also update `cargo-hakari`, `cargo-deny`, `cargo-hack` and
`cargo-nextest` to their latest versions.

Prior update was in #9445.
2024-12-05 13:00:40 +02:00
Erik Grinaker
2fa461b668 Makefile: build pg_visibility (#9922)
Build the `pg_visibility` extension for use with `neon_local`. This is
useful to inspect the visibility map for debugging.

Touches #9914.
2024-12-05 13:00:40 +02:00
Vlad Lazar
03d90bc0b3 remote_storage/abs: count 404 and 304 for get as ok for metrics (#9912)
## Problem

We currently see elevated levels of errors for GetBlob requests. This is
because 404 and 304 are counted as errors for metric reporting.

## Summary of Changes

Bring the implementation in line with the S3 client and treat 404 and
304 responses as ok for metric purposes.

Related: https://github.com/neondatabase/cloud/issues/20666
2024-12-05 13:00:40 +02:00
Ivan Efremov
268bc890ea proxy: spawn cancellation checks in the background (#9918)
## Problem
For cancellation, a connection is open during all the cancel checks.
## Summary of changes
Spawn cancellation checks in the background, and close connection
immediately.
Use task_tracker for cancellation checks.
2024-12-05 13:00:40 +02:00
Folke Behrens
8a6ee79f6f Merge pull request #9921 from neondatabase/rc/release-proxy/2024-11-28
Proxy release 2024-11-28
2024-11-28 11:09:06 +01:00
github-actions[bot]
9052c32b46 Proxy release 2024-11-28 2024-11-28 06:02:15 +00:00
Ivan Efremov
995e729ebe Merge pull request #9832 from neondatabase/rc/release-proxy/2024-11-21
Proxy release 2024-11-21
2024-11-21 09:41:31 +02:00
github-actions[bot]
76077e1ddf Proxy release 2024-11-21 2024-11-21 06:02:11 +00:00
Ivan Efremov
0467d88f06 Merge pull request #9756 from neondatabase/rc/proxy/2024-11-14
Proxy release 2024-11-14
2024-11-14 09:46:52 +02:00
Ivan Efremov
f5eec194e7 Merge pull request #9674 from neondatabase/rc/proxy/2024-11-07
Proxy release 2024-11-07
2024-11-07 12:07:12 +02:00
Conrad Ludgate
7e00be391d Merge pull request #9558 from neondatabase/rc/proxy/2024-10-29
Auth broker release 2024-10-29
2024-10-29 12:10:50 +00:00
Folke Behrens
d56599df2a Merge pull request #9499 from neondatabase/rc/proxy/2024-10-24
Proxy release 2024-10-24
2024-10-24 10:34:56 +02:00
Folke Behrens
9d9aab3680 Merge pull request #9426 from neondatabase/rc/proxy/2024-10-17
Proxy release 2024-10-17
2024-10-17 12:18:51 +02:00
Folke Behrens
a202b1b5cc Merge pull request #9341 from neondatabase/rc/proxy/2024-10-10
Proxy release 2024-10-10
2024-10-10 09:17:11 +02:00
Folke Behrens
90f731f3b1 Merge pull request #9256 from neondatabase/rc/proxy/2024-10-03
Proxy release 2024-10-03
2024-10-03 11:01:41 +02:00
Conrad Ludgate
7736b748d3 Merge pull request #9159 from neondatabase/rc/proxy/2024-09-26
Proxy release 2024-09-26
2024-09-26 09:22:33 +01:00
Conrad Ludgate
9c23333cb3 Merge pull request #9056 from neondatabase/rc/proxy/2024-09-19
Proxy release 2024-09-19
2024-09-19 10:41:17 +01:00
Conrad Ludgate
66a99009ba Merge pull request #8799 from neondatabase/rc/proxy/2024-08-22
Proxy release 2024-08-22
2024-08-22 10:04:56 +01:00
Conrad Ludgate
5d4c57491f Merge pull request #8723 from neondatabase/rc/proxy/2024-08-14
Proxy release 2024-08-14
2024-08-14 13:05:51 +01:00
Conrad Ludgate
73935ea3a2 Merge pull request #8647 from neondatabase/rc/proxy/2024-08-08
Proxy release 2024-08-08
2024-08-08 15:37:09 +01:00
Conrad Ludgate
32e595d4dd Merge branch 'release-proxy' into rc/proxy/2024-08-08 2024-08-08 13:53:33 +01:00
Conrad Ludgate
b0d69acb07 Merge pull request #8505 from neondatabase/rc/proxy/2024-07-25
Proxy release 2024-07-25
2024-07-25 11:07:19 +01:00
Conrad Ludgate
98355a419a Merge pull request #8351 from neondatabase/rc/proxy/2024-07-11
Proxy release 2024-07-11
2024-07-11 10:40:17 +01:00
Conrad Ludgate
cfb03d6cf0 Merge pull request #8178 from neondatabase/rc/proxy/2024-06-27
Proxy release 2024-06-27
2024-06-27 11:35:30 +01:00
Conrad Ludgate
d81ef3f962 Revert "proxy: update tokio-postgres to allow arbitrary config params (#8076)"
This reverts commit 78d9059fc7.
2024-06-27 09:46:58 +01:00
Conrad Ludgate
5d62c67e75 Merge pull request #8117 from neondatabase/rc/proxy/2024-06-20
Proxy release 2024-06-20
2024-06-20 11:42:35 +01:00
Anna Khanova
53d53d5b1e Merge pull request #7980 from neondatabase/rc/proxy/2024-06-06
Proxy release 2024-06-06
2024-06-06 13:14:40 +02:00
Anna Khanova
29fe6ea47a Merge pull request #7909 from neondatabase/rc/proxy/2024-05-30
Proxy release 2024-05-30
2024-05-30 14:59:41 +02:00
Alexander Bayandin
640327ccb3 Merge pull request #7880 from neondatabase/rc/proxy/2024-05-24
Proxy release 2024-05-24
2024-05-24 18:00:18 +01:00
Anna Khanova
7cf0f6b37e Merge pull request #7853 from neondatabase/rc/proxy/2024-05-23
Proxy release 2024-05-23
2024-05-23 12:09:13 +02:00
Anna Khanova
03c2c569be [proxy] Do not fail after parquet upload error (#7858)
## Problem

If the parquet upload was unsuccessful, it will panic.

## Summary of changes

Write error in logs instead.
2024-05-23 11:44:47 +02:00
Conrad Ludgate
eff6d4538a Merge pull request #7654 from neondatabase/rc/proxy/2024-05-08
Proxy release 2024-05-08
2024-05-08 11:56:20 +01:00
Conrad Ludgate
5ef7782e9c Merge pull request #7649 from neondatabase/rc/proxy/2024-05-08
Proxy release 2024-05-08
2024-05-08 06:54:03 +01:00
Conrad Ludgate
73101db8c4 Merge branch 'release-proxy' into rc/proxy/2024-05-08 2024-05-08 06:43:57 +01:00
Anna Khanova
bccdfc6d39 Merge pull request #7580 from neondatabase/rc/proxy/2024-05-02
Proxy release 2024-05-02
2024-05-02 12:00:01 +02:00
Anna Khanova
99595813bb proxy: keep track on the number of events from redis by type. (#7582)
## Problem

It's unclear what is the distribution of messages, proxy is consuming
from redis.

## Summary of changes

Add counter.
2024-05-02 11:56:19 +02:00
Anna Khanova
fe07b54758 Merge pull request #7507 from neondatabase/rc/proxy/2024-04-25
Proxy release 2024-04-25
2024-04-25 13:50:05 +02:00
Anna Khanova
a42d173e7b proxy: Fix cancellations (#7510)
## Problem

Cancellations were published to the channel, that was never read.

## Summary of changes

Fallback to global redis publishing.
2024-04-25 13:42:25 +02:00
Anna Khanova
e07f689238 Update connect to compute and wake compute retry configs (#7509)
## Problem

## Summary of changes

Decrease waiting time
2024-04-25 13:20:21 +02:00
Conrad Ludgate
7831eddc88 Merge pull request #7417 from neondatabase/rc/proxy/2024-04-18
Proxy release 2024-04-18
2024-04-18 12:03:07 +01:00
Conrad Ludgate
943b1bc80c Merge pull request #7366 from neondatabase/proxy-hotfix
Release proxy (2024-04-11 hotfix)
2024-04-12 10:15:14 +01:00
Conrad Ludgate
95a184e9b7 proxy: fix overloaded db connection closure (#7364)
## Problem

possible for the database connections to not close in time.

## Summary of changes

force the closing of connections if the client has hung up
2024-04-11 23:38:47 +01:00
Conrad Ludgate
3fa17e9d17 Merge pull request #7357 from neondatabase/rc/proxy/2024-04-11
Proxy release 2024-04-11
2024-04-11 11:49:45 +01:00
Anna Khanova
55e0fd9789 Merge pull request #7304 from neondatabase/rc/proxy/2024-04-04
Proxy release 2024-04-04
2024-04-04 12:40:11 +02:00
Anna Khanova
2a88889f44 Merge pull request #7254 from neondatabase/rc/proxy/2024-03-27
Proxy release 2024-03-27
2024-03-27 11:44:09 +01:00
Conrad Ludgate
5bad8126dc Merge pull request #7173 from neondatabase/rc/proxy/2024-03-19
Proxy release 2024-03-19
2024-03-19 12:11:42 +00:00
Anna Khanova
27bc242085 Merge pull request #7119 from neondatabase/rc/proxy/2024-03-14
Proxy release 2024-03-14
2024-03-14 14:57:05 +05:00
Anna Khanova
192b49cc6d Merge branch 'release-proxy' into rc/proxy/2024-03-14 2024-03-14 14:16:36 +05:00
Conrad Ludgate
e1b60f3693 Merge pull request #7041 from neondatabase/rc/proxy/2024-03-07
Proxy release 2024-03-07
2024-03-08 08:19:16 +00:00
Anna Khanova
2804f5323b Merge pull request #6997 from neondatabase/rc/proxy/2024-03-04
Proxy release 2024-03-04
2024-03-04 17:36:11 +04:00
Anna Khanova
676adc6b32 Merge branch 'release-proxy' into rc/proxy/2024-03-04 2024-03-04 16:41:46 +04:00
194 changed files with 2688 additions and 9106 deletions

View File

@@ -49,6 +49,10 @@ inputs:
description: 'A JSON object with project settings'
required: false
default: '{}'
default_endpoint_settings:
description: 'A JSON object with the default endpoint settings'
required: false
default: '{}'
outputs:
dsn:
@@ -135,6 +139,21 @@ runs:
-H "Accept: application/json" -H "Content-Type: application/json" -H "Authorization: Bearer ${ADMIN_API_KEY}" \
-d "{\"scheduling\": \"Essential\"}"
fi
# XXX
# This is a workaround for the default endpoint settings, which currently do not allow some settings in the public API.
# https://github.com/neondatabase/cloud/issues/27108
if [[ -n ${DEFAULT_ENDPOINT_SETTINGS} && ${DEFAULT_ENDPOINT_SETTINGS} != "{}" ]] ; then
PROJECT_DATA=$(curl -X GET \
"https://${API_HOST}/regions/${REGION_ID}/api/v1/admin/projects/${project_id}" \
-H "Accept: application/json" -H "Content-Type: application/json" -H "Authorization: Bearer ${ADMIN_API_KEY}" \
-d "{\"scheduling\": \"Essential\"}"
)
NEW_DEFAULT_ENDPOINT_SETTINGS=$(echo ${PROJECT_DATA} | jq -rc ".project.default_endpoint_settings + ${DEFAULT_ENDPOINT_SETTINGS}")
curl -X POST --fail \
"https://${API_HOST}/regions/${REGION_ID}/api/v1/admin/projects/${project_id}/default_endpoint_settings" \
-H "Accept: application/json" -H "Content-Type: application/json" -H "Authorization: Bearer ${ADMIN_API_KEY}" \
--data "${NEW_DEFAULT_ENDPOINT_SETTINGS}"
fi
env:
@@ -152,3 +171,4 @@ runs:
PSQL: ${{ inputs.psql_path }}
LD_LIBRARY_PATH: ${{ inputs.libpq_lib_path }}
PROJECT_SETTINGS: ${{ inputs.project_settings }}
DEFAULT_ENDPOINT_SETTINGS: ${{ inputs.default_endpoint_settings }}

View File

@@ -279,14 +279,18 @@ jobs:
# run all non-pageserver tests
${cov_prefix} cargo nextest run $CARGO_FLAGS $CARGO_FEATURES -E '!package(pageserver)'
# run pageserver tests
# (When developing new pageserver features gated by config fields, we commonly make the rust
# unit tests sensitive to an environment variable NEON_PAGESERVER_UNIT_TEST_FEATURENAME.
# Then run the nextest invocation below for all relevant combinations. Singling out the
# pageserver tests from non-pageserver tests cuts down the time it takes for this CI step.)
NEON_PAGESERVER_UNIT_TEST_VIRTUAL_FILE_IOENGINE=tokio-epoll-uring \
${cov_prefix} \
cargo nextest run $CARGO_FLAGS $CARGO_FEATURES -E 'package(pageserver)'
# run pageserver tests with different settings
for get_vectored_concurrent_io in sequential sidecar-task; do
for io_engine in std-fs tokio-epoll-uring ; do
for io_mode in buffered direct direct-rw ; do
NEON_PAGESERVER_UNIT_TEST_GET_VECTORED_CONCURRENT_IO=$get_vectored_concurrent_io \
NEON_PAGESERVER_UNIT_TEST_VIRTUAL_FILE_IOENGINE=$io_engine \
NEON_PAGESERVER_UNIT_TEST_VIRTUAL_FILE_IO_MODE=$io_mode \
${cov_prefix} \
cargo nextest run $CARGO_FLAGS $CARGO_FEATURES -E 'package(pageserver)'
done
done
done
# Run separate tests for real S3
export ENABLE_REAL_S3_REMOTE_STORAGE=nonempty
@@ -401,6 +405,8 @@ jobs:
CHECK_ONDISK_DATA_COMPATIBILITY: nonempty
BUILD_TAG: ${{ inputs.build-tag }}
PAGESERVER_VIRTUAL_FILE_IO_ENGINE: tokio-epoll-uring
PAGESERVER_GET_VECTORED_CONCURRENT_IO: sidecar-task
PAGESERVER_VIRTUAL_FILE_IO_MODE: direct-rw
USE_LFC: ${{ matrix.lfc_state == 'with-lfc' && 'true' || 'false' }}
# Temporary disable this step until we figure out why it's so flaky

View File

@@ -314,8 +314,7 @@ jobs:
test_selection: performance
run_in_parallel: false
save_perf_report: ${{ github.ref_name == 'main' }}
# test_pageserver_max_throughput_getpage_at_latest_lsn is run in separate workflow periodic_pagebench.yml because it needs snapshots
extra_params: --splits 5 --group ${{ matrix.pytest_split_group }} --ignore=test_runner/performance/pageserver/pagebench/test_pageserver_max_throughput_getpage_at_latest_lsn.py
extra_params: --splits 5 --group ${{ matrix.pytest_split_group }}
benchmark_durations: ${{ needs.get-benchmarks-durations.outputs.json }}
pg_version: v16
aws-oidc-role-arn: ${{ vars.DEV_AWS_OIDC_ROLE_ARN }}
@@ -324,6 +323,8 @@ jobs:
PERF_TEST_RESULT_CONNSTR: "${{ secrets.PERF_TEST_RESULT_CONNSTR }}"
TEST_RESULT_CONNSTR: "${{ secrets.REGRESS_TEST_RESULT_CONNSTR_NEW }}"
PAGESERVER_VIRTUAL_FILE_IO_ENGINE: tokio-epoll-uring
PAGESERVER_GET_VECTORED_CONCURRENT_IO: sidecar-task
PAGESERVER_VIRTUAL_FILE_IO_MODE: direct-rw
SYNC_BETWEEN_TESTS: true
# XXX: no coverage data handling here, since benchmarks are run on release builds,
# while coverage is currently collected for the debug ones
@@ -964,7 +965,7 @@ jobs:
fi
- name: Verify docker-compose example and test extensions
timeout-minutes: 60
timeout-minutes: 20
env:
TAG: >-
${{

View File

@@ -35,7 +35,7 @@ jobs:
matrix:
pg-version: [16, 17]
runs-on: us-east-2
runs-on: [ self-hosted, small ]
container:
# We use the neon-test-extensions image here as it contains the source code for the extensions.
image: ghcr.io/neondatabase/neon-test-extensions-v${{ matrix.pg-version }}:latest
@@ -71,7 +71,20 @@ jobs:
region_id: ${{ inputs.region_id || 'aws-us-east-2' }}
postgres_version: ${{ matrix.pg-version }}
project_settings: ${{ steps.project-settings.outputs.settings }}
# We need these settings to get the expected output results.
# We cannot use the environment variables e.g. PGTZ due to
# https://github.com/neondatabase/neon/issues/1287
default_endpoint_settings: >
{
"pg_settings": {
"DateStyle": "Postgres,MDY",
"TimeZone": "America/Los_Angeles",
"compute_query_id": "off",
"neon.allow_unstable_extensions": "on"
}
}
api_key: ${{ secrets.NEON_STAGING_API_KEY }}
admin_api_key: ${{ secrets.NEON_STAGING_ADMIN_API_KEY }}
- name: Run the regression tests
run: /run-tests.sh -r /ext-src

View File

@@ -63,10 +63,8 @@ jobs:
- name: Filter out only v-string for build matrix
id: postgres_changes
env:
CHANGES: ${{ steps.files_changed.outputs.changes }}
run: |
v_strings_only_as_json_array=$(echo ${CHANGES} | jq '.[]|select(test("v\\d+"))' | jq --slurp -c)
v_strings_only_as_json_array=$(echo ${{ steps.files_changed.outputs.chnages }} | jq '.[]|select(test("v\\d+"))' | jq --slurp -c)
echo "changes=${v_strings_only_as_json_array}" | tee -a "${GITHUB_OUTPUT}"
check-macos-build:

View File

@@ -1,4 +1,4 @@
name: Periodic pagebench performance test on unit-perf hetzner runner
name: Periodic pagebench performance test on dedicated EC2 machine in eu-central-1 region
on:
schedule:
@@ -8,7 +8,7 @@ on:
# │ │ ┌───────────── day of the month (1 - 31)
# │ │ │ ┌───────────── month (1 - 12 or JAN-DEC)
# │ │ │ │ ┌───────────── day of the week (0 - 6 or SUN-SAT)
- cron: '0 */4 * * *' # Runs every 4 hours
- cron: '0 */3 * * *' # Runs every 3 hours
workflow_dispatch: # Allows manual triggering of the workflow
inputs:
commit_hash:
@@ -16,11 +16,6 @@ on:
description: 'The long neon repo commit hash for the system under test (pageserver) to be tested.'
required: false
default: ''
recreate_snapshots:
type: boolean
description: 'Recreate snapshots - !!!WARNING!!! We should only recreate snapshots if the previous ones are no longer compatible. Otherwise benchmarking results are not comparable across runs.'
required: false
default: false
defaults:
run:
@@ -34,13 +29,13 @@ permissions:
contents: read
jobs:
run_periodic_pagebench_test:
trigger_bench_on_ec2_machine_in_eu_central_1:
permissions:
id-token: write # aws-actions/configure-aws-credentials
statuses: write
contents: write
pull-requests: write
runs-on: [ self-hosted, unit-perf ]
runs-on: [ self-hosted, small ]
container:
image: ghcr.io/neondatabase/build-tools:pinned-bookworm
credentials:
@@ -49,13 +44,10 @@ jobs:
options: --init
timeout-minutes: 360 # Set the timeout to 6 hours
env:
API_KEY: ${{ secrets.PERIODIC_PAGEBENCH_EC2_RUNNER_API_KEY }}
RUN_ID: ${{ github.run_id }}
DEFAULT_PG_VERSION: 16
BUILD_TYPE: release
RUST_BACKTRACE: 1
# NEON_ENV_BUILDER_USE_OVERLAYFS_FOR_SNAPSHOTS: 1 - doesn't work without root in container
S3_BUCKET: neon-github-public-dev
PERF_TEST_RESULT_CONNSTR: "${{ secrets.PERF_TEST_RESULT_CONNSTR }}"
AWS_DEFAULT_REGION : "eu-central-1"
AWS_INSTANCE_ID : "i-02a59a3bf86bc7e74"
steps:
# we don't need the neon source code because we run everything remotely
# however we still need the local github actions to run the allure step below
@@ -64,194 +56,99 @@ jobs:
with:
egress-policy: audit
- name: Set up the environment which depends on $RUNNER_TEMP on nvme drive
id: set-env
shell: bash -euxo pipefail {0}
run: |
{
echo "NEON_DIR=${RUNNER_TEMP}/neon"
echo "NEON_BIN=${RUNNER_TEMP}/neon/bin"
echo "POSTGRES_DISTRIB_DIR=${RUNNER_TEMP}/neon/pg_install"
echo "LD_LIBRARY_PATH=${RUNNER_TEMP}/neon/pg_install/v${DEFAULT_PG_VERSION}/lib"
echo "BACKUP_DIR=${RUNNER_TEMP}/instance_store/saved_snapshots"
echo "TEST_OUTPUT=${RUNNER_TEMP}/neon/test_output"
echo "PERF_REPORT_DIR=${RUNNER_TEMP}/neon/test_output/perf-report-local"
echo "ALLURE_DIR=${RUNNER_TEMP}/neon/test_output/allure-results"
echo "ALLURE_RESULTS_DIR=${RUNNER_TEMP}/neon/test_output/allure-results/results"
} >> "$GITHUB_ENV"
- uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
echo "allure_results_dir=${RUNNER_TEMP}/neon/test_output/allure-results/results" >> "$GITHUB_OUTPUT"
- name: Show my own (github runner) external IP address - usefull for IP allowlisting
run: curl https://ifconfig.me
- uses: aws-actions/configure-aws-credentials@e3dd6a429d7300a6a4c196c26e071d42e0343502 # v4.0.2
- name: Assume AWS OIDC role that allows to manage (start/stop/describe... EC machine)
uses: aws-actions/configure-aws-credentials@e3dd6a429d7300a6a4c196c26e071d42e0343502 # v4.0.2
with:
aws-region: eu-central-1
role-to-assume: ${{ vars.DEV_AWS_OIDC_ROLE_ARN }}
role-duration-seconds: 18000 # max 5 hours (needed in case commit hash is still being built)
role-to-assume: ${{ vars.DEV_AWS_OIDC_ROLE_MANAGE_BENCHMARK_EC2_VMS_ARN }}
role-duration-seconds: 3600
- name: Start EC2 instance and wait for the instance to boot up
run: |
aws ec2 start-instances --instance-ids $AWS_INSTANCE_ID
aws ec2 wait instance-running --instance-ids $AWS_INSTANCE_ID
sleep 60 # sleep some time to allow cloudinit and our API server to start up
- name: Determine public IP of the EC2 instance and set env variable EC2_MACHINE_URL_US
run: |
public_ip=$(aws ec2 describe-instances --instance-ids $AWS_INSTANCE_ID --query 'Reservations[*].Instances[*].PublicIpAddress' --output text)
echo "Public IP of the EC2 instance: $public_ip"
echo "EC2_MACHINE_URL_US=https://${public_ip}:8443" >> $GITHUB_ENV
- name: Determine commit hash
id: commit_hash
shell: bash -euxo pipefail {0}
env:
INPUT_COMMIT_HASH: ${{ github.event.inputs.commit_hash }}
run: |
if [[ -z "${INPUT_COMMIT_HASH}" ]]; then
COMMIT_HASH=$(curl -s https://api.github.com/repos/neondatabase/neon/commits/main | jq -r '.sha')
echo "COMMIT_HASH=$COMMIT_HASH" >> $GITHUB_ENV
echo "commit_hash=$COMMIT_HASH" >> "$GITHUB_OUTPUT"
if [ -z "$INPUT_COMMIT_HASH" ]; then
echo "COMMIT_HASH=$(curl -s https://api.github.com/repos/neondatabase/neon/commits/main | jq -r '.sha')" >> $GITHUB_ENV
echo "COMMIT_HASH_TYPE=latest" >> $GITHUB_ENV
else
COMMIT_HASH="${INPUT_COMMIT_HASH}"
echo "COMMIT_HASH=$COMMIT_HASH" >> $GITHUB_ENV
echo "commit_hash=$COMMIT_HASH" >> "$GITHUB_OUTPUT"
echo "COMMIT_HASH=$INPUT_COMMIT_HASH" >> $GITHUB_ENV
echo "COMMIT_HASH_TYPE=manual" >> $GITHUB_ENV
fi
- name: Checkout the neon repository at given commit hash
uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
with:
ref: ${{ steps.commit_hash.outputs.commit_hash }}
# does not reuse ./.github/actions/download because we need to download the artifact for the given commit hash
# example artifact
# s3://neon-github-public-dev/artifacts/48b870bc078bd2c450eb7b468e743b9c118549bf/15036827400/1/neon-Linux-X64-release-artifact.tar.zst /instance_store/artifacts/neon-Linux-release-artifact.tar.zst
- name: Determine artifact S3_KEY for given commit hash and download and extract artifact
id: artifact_prefix
shell: bash -euxo pipefail {0}
env:
ARCHIVE: ${{ runner.temp }}/downloads/neon-${{ runner.os }}-${{ runner.arch }}-release-artifact.tar.zst
COMMIT_HASH: ${{ env.COMMIT_HASH }}
COMMIT_HASH_TYPE: ${{ env.COMMIT_HASH_TYPE }}
- name: Start Bench with run_id
run: |
attempt=0
max_attempts=24 # 5 minutes * 24 = 2 hours
curl -k -X 'POST' \
"${EC2_MACHINE_URL_US}/start_test/${GITHUB_RUN_ID}" \
-H 'accept: application/json' \
-H 'Content-Type: application/json' \
-H "Authorization: Bearer $API_KEY" \
-d "{\"neonRepoCommitHash\": \"${COMMIT_HASH}\", \"neonRepoCommitHashType\": \"${COMMIT_HASH_TYPE}\"}"
while [[ $attempt -lt $max_attempts ]]; do
# the following command will fail until the artifacts are available ...
S3_KEY=$(aws s3api list-objects-v2 --bucket "$S3_BUCKET" --prefix "artifacts/$COMMIT_HASH/" \
| jq -r '.Contents[]?.Key' \
| grep "neon-${{ runner.os }}-${{ runner.arch }}-release-artifact.tar.zst" \
| sort --version-sort \
| tail -1) || true # ... thus ignore errors from the command
if [[ -n "${S3_KEY}" ]]; then
echo "Artifact found: $S3_KEY"
echo "S3_KEY=$S3_KEY" >> $GITHUB_ENV
- name: Poll Test Status
id: poll_step
run: |
status=""
while [[ "$status" != "failure" && "$status" != "success" ]]; do
response=$(curl -k -X 'GET' \
"${EC2_MACHINE_URL_US}/test_status/${GITHUB_RUN_ID}" \
-H 'accept: application/json' \
-H "Authorization: Bearer $API_KEY")
echo "Response: $response"
set +x
status=$(echo $response | jq -r '.status')
echo "Test status: $status"
if [[ "$status" == "failure" ]]; then
echo "Test failed"
exit 1 # Fail the job step if status is failure
elif [[ "$status" == "success" || "$status" == "null" ]]; then
break
elif [[ "$status" == "too_many_runs" ]]; then
echo "Too many runs already running"
echo "too_many_runs=true" >> "$GITHUB_OUTPUT"
exit 1
fi
# Increment attempt counter and sleep for 5 minutes
attempt=$((attempt + 1))
echo "Attempt $attempt of $max_attempts to find artifacts in S3 bucket s3://$S3_BUCKET/artifacts/$COMMIT_HASH failed. Retrying in 5 minutes..."
sleep 300 # Sleep for 5 minutes
sleep 60 # Poll every 60 seconds
done
if [[ -z "${S3_KEY}" ]]; then
echo "Error: artifact not found in S3 bucket s3://$S3_BUCKET/artifacts/$COMMIT_HASH" after 2 hours
else
mkdir -p $(dirname $ARCHIVE)
time aws s3 cp --only-show-errors s3://$S3_BUCKET/${S3_KEY} ${ARCHIVE}
mkdir -p ${NEON_DIR}
time tar -xf ${ARCHIVE} -C ${NEON_DIR}
rm -f ${ARCHIVE}
fi
- name: Download snapshots from S3
if: ${{ github.event_name != 'workflow_dispatch' || github.event.inputs.recreate_snapshots == 'false' || github.event.inputs.recreate_snapshots == '' }}
id: download_snapshots
shell: bash -euxo pipefail {0}
- name: Retrieve Test Logs
if: always() && steps.poll_step.outputs.too_many_runs != 'true'
run: |
# Download the snapshots from S3
mkdir -p ${TEST_OUTPUT}
mkdir -p $BACKUP_DIR
cd $BACKUP_DIR
mkdir parts
cd parts
PART=$(aws s3api list-objects-v2 --bucket $S3_BUCKET --prefix performance/pagebench/ \
| jq -r '.Contents[]?.Key' \
| grep -E 'shared-snapshots-[0-9]{4}-[0-9]{2}-[0-9]{2}' \
| sort \
| tail -1)
echo "Latest PART: $PART"
if [[ -z "$PART" ]]; then
echo "ERROR: No matching S3 key found" >&2
exit 1
fi
S3_KEY=$(dirname $PART)
time aws s3 cp --only-show-errors --recursive s3://${S3_BUCKET}/$S3_KEY/ .
cd $TEST_OUTPUT
time cat $BACKUP_DIR/parts/* | zstdcat | tar --extract --preserve-permissions
rm -rf ${BACKUP_DIR}
curl -k -X 'GET' \
"${EC2_MACHINE_URL_US}/test_log/${GITHUB_RUN_ID}" \
-H 'accept: application/gzip' \
-H "Authorization: Bearer $API_KEY" \
--output "test_log_${GITHUB_RUN_ID}.gz"
- name: Cache poetry deps
uses: actions/cache@v4
with:
path: ~/.cache/pypoetry/virtualenvs
key: v2-${{ runner.os }}-${{ runner.arch }}-python-deps-bookworm-${{ hashFiles('poetry.lock') }}
- name: Install Python deps
shell: bash -euxo pipefail {0}
run: ./scripts/pysync
# we need high number of open files for pagebench
- name: show ulimits
shell: bash -euxo pipefail {0}
- name: Unzip Test Log and Print it into this job's log
if: always() && steps.poll_step.outputs.too_many_runs != 'true'
run: |
ulimit -a
- name: Run pagebench testcase
shell: bash -euxo pipefail {0}
env:
CI: false # need to override this env variable set by github to enforce using snapshots
run: |
export PLATFORM=hetzner-unit-perf-${COMMIT_HASH_TYPE}
# report the commit hash of the neon repository in the revision of the test results
export GITHUB_SHA=${COMMIT_HASH}
rm -rf ${PERF_REPORT_DIR}
rm -rf ${ALLURE_RESULTS_DIR}
mkdir -p ${PERF_REPORT_DIR}
mkdir -p ${ALLURE_RESULTS_DIR}
PARAMS="--alluredir=${ALLURE_RESULTS_DIR} --tb=short --verbose -rA"
EXTRA_PARAMS="--out-dir ${PERF_REPORT_DIR} --durations-path $TEST_OUTPUT/benchmark_durations.json"
# run only two selected tests
# environment set by parent:
# RUST_BACKTRACE=1 DEFAULT_PG_VERSION=16 BUILD_TYPE=release
./scripts/pytest ${PARAMS} test_runner/performance/pageserver/pagebench/test_pageserver_max_throughput_getpage_at_latest_lsn.py::test_pageserver_characterize_throughput_with_n_tenants ${EXTRA_PARAMS}
./scripts/pytest ${PARAMS} test_runner/performance/pageserver/pagebench/test_pageserver_max_throughput_getpage_at_latest_lsn.py::test_pageserver_characterize_latencies_with_1_client_and_throughput_with_many_clients_one_tenant ${EXTRA_PARAMS}
- name: upload the performance metrics to the Neon performance database which is used by grafana dashboards to display the results
shell: bash -euxo pipefail {0}
run: |
export REPORT_FROM="$PERF_REPORT_DIR"
export GITHUB_SHA=${COMMIT_HASH}
time ./scripts/generate_and_push_perf_report.sh
- name: Upload test results
if: ${{ !cancelled() }}
uses: ./.github/actions/allure-report-store
with:
report-dir: ${{ steps.set-env.outputs.allure_results_dir }}
unique-key: ${{ env.BUILD_TYPE }}-${{ env.DEFAULT_PG_VERSION }}-${{ runner.arch }}
aws-oidc-role-arn: ${{ vars.DEV_AWS_OIDC_ROLE_ARN }}
gzip -d "test_log_${GITHUB_RUN_ID}.gz"
cat "test_log_${GITHUB_RUN_ID}"
- name: Create Allure report
id: create-allure-report
if: ${{ !cancelled() }}
uses: ./.github/actions/allure-report-generate
with:
aws-oidc-role-arn: ${{ vars.DEV_AWS_OIDC_ROLE_ARN }}
- name: Upload snapshots
if: ${{ github.event_name == 'workflow_dispatch' && github.event.inputs.recreate_snapshots != 'false' && github.event.inputs.recreate_snapshots != '' }}
id: upload_snapshots
shell: bash -euxo pipefail {0}
run: |
mkdir -p $BACKUP_DIR
cd $TEST_OUTPUT
tar --create --preserve-permissions --file - shared-snapshots | zstd -o $BACKUP_DIR/shared_snapshots.tar.zst
cd $BACKUP_DIR
mkdir parts
split -b 1G shared_snapshots.tar.zst ./parts/shared_snapshots.tar.zst.part.
SNAPSHOT_DATE=$(date +%F) # YYYY-MM-DD
cd parts
time aws s3 cp --recursive . s3://${S3_BUCKET}/performance/pagebench/shared-snapshots-${SNAPSHOT_DATE}/
- name: Post to a Slack channel
if: ${{ github.event.schedule && failure() }}
uses: slackapi/slack-github-action@fcfb566f8b0aab22203f066d80ca1d7e4b5d05b3 # v1.27.1
@@ -260,22 +157,26 @@ jobs:
slack-message: "Periodic pagebench testing on dedicated hardware: ${{ job.status }}\n${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}"
env:
SLACK_BOT_TOKEN: ${{ secrets.SLACK_BOT_TOKEN }}
- name: Cleanup Test Resources
if: always()
shell: bash -euxo pipefail {0}
env:
ARCHIVE: ${{ runner.temp }}/downloads/neon-${{ runner.os }}-${{ runner.arch }}-release-artifact.tar.zst
run: |
# Cleanup the test resources
if [[ -d "${BACKUP_DIR}" ]]; then
rm -rf ${BACKUP_DIR}
fi
if [[ -d "${TEST_OUTPUT}" ]]; then
rm -rf ${TEST_OUTPUT}
fi
if [[ -d "${NEON_DIR}" ]]; then
rm -rf ${NEON_DIR}
fi
rm -rf $(dirname $ARCHIVE)
curl -k -X 'POST' \
"${EC2_MACHINE_URL_US}/cleanup_test/${GITHUB_RUN_ID}" \
-H 'accept: application/json' \
-H "Authorization: Bearer $API_KEY" \
-d ''
- name: Assume AWS OIDC role that allows to manage (start/stop/describe... EC machine)
if: always() && steps.poll_step.outputs.too_many_runs != 'true'
uses: aws-actions/configure-aws-credentials@e3dd6a429d7300a6a4c196c26e071d42e0343502 # v4.0.2
with:
aws-region: eu-central-1
role-to-assume: ${{ vars.DEV_AWS_OIDC_ROLE_MANAGE_BENCHMARK_EC2_VMS_ARN }}
role-duration-seconds: 3600
- name: Stop EC2 instance and wait for the instance to be stopped
if: always() && steps.poll_step.outputs.too_many_runs != 'true'
run: |
aws ec2 stop-instances --instance-ids $AWS_INSTANCE_ID
aws ec2 wait instance-stopped --instance-ids $AWS_INSTANCE_ID

294
Cargo.lock generated
View File

@@ -1112,12 +1112,6 @@ version = "1.0.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "baf1de4339761588bc0619e3cbc0120ee582ebb74b53b4efbf79117bd2da40fd"
[[package]]
name = "cfg_aliases"
version = "0.2.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "613afe47fcd5fac7ccf1db93babcb082c5994d996f20b8b159f2ad1658eb5724"
[[package]]
name = "cgroups-rs"
version = "0.3.3"
@@ -1276,7 +1270,7 @@ version = "0.1.0"
dependencies = [
"anyhow",
"chrono",
"indexmap 2.9.0",
"indexmap 2.0.1",
"jsonwebtoken",
"regex",
"remote_storage",
@@ -1308,11 +1302,11 @@ dependencies = [
"flate2",
"futures",
"http 1.1.0",
"indexmap 2.9.0",
"indexmap 2.0.1",
"itertools 0.10.5",
"jsonwebtoken",
"metrics",
"nix 0.30.1",
"nix 0.27.1",
"notify",
"num_cpus",
"once_cell",
@@ -1435,7 +1429,7 @@ dependencies = [
"humantime-serde",
"hyper 0.14.30",
"jsonwebtoken",
"nix 0.30.1",
"nix 0.27.1",
"once_cell",
"pageserver_api",
"pageserver_client",
@@ -2597,7 +2591,7 @@ dependencies = [
"futures-sink",
"futures-util",
"http 0.2.9",
"indexmap 2.9.0",
"indexmap 2.0.1",
"slab",
"tokio",
"tokio-util",
@@ -2616,7 +2610,7 @@ dependencies = [
"futures-sink",
"futures-util",
"http 1.1.0",
"indexmap 2.9.0",
"indexmap 2.0.1",
"slab",
"tokio",
"tokio-util",
@@ -2863,14 +2857,14 @@ dependencies = [
"pprof",
"regex",
"routerify",
"rustls 0.23.27",
"rustls 0.23.18",
"rustls-pemfile 2.1.1",
"serde",
"serde_json",
"serde_path_to_error",
"thiserror 1.0.69",
"tokio",
"tokio-rustls 0.26.2",
"tokio-rustls 0.26.0",
"tokio-stream",
"tokio-util",
"tracing",
@@ -3200,12 +3194,12 @@ dependencies = [
[[package]]
name = "indexmap"
version = "2.9.0"
version = "2.0.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "cea70ddb795996207ad57735b50c5982d8844f38ba9ee5f1aedcfb708a2aa11e"
checksum = "ad227c3af19d4914570ad36d30409928b75967c298feb9ea1969db3a610bb14e"
dependencies = [
"equivalent",
"hashbrown 0.15.2",
"hashbrown 0.14.5",
"serde",
]
@@ -3228,7 +3222,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "232929e1d75fe899576a3d5c7416ad0d88dbfbb3c3d6aa00873a7408a50ddb88"
dependencies = [
"ahash",
"indexmap 2.9.0",
"indexmap 2.0.1",
"is-terminal",
"itoa",
"log",
@@ -3251,7 +3245,7 @@ dependencies = [
"crossbeam-utils",
"dashmap 6.1.0",
"env_logger",
"indexmap 2.9.0",
"indexmap 2.0.1",
"itoa",
"log",
"num-format",
@@ -3518,9 +3512,9 @@ dependencies = [
[[package]]
name = "libc"
version = "0.2.172"
version = "0.2.169"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d750af042f7ef4f724306de029d18836c26c1765a54a6a3f094cbd23a7267ffa"
checksum = "b5aba8db14291edd000dfcc4d620c7ebfb122c613afb886ca8803fa4e128a20a"
[[package]]
name = "libloading"
@@ -3794,16 +3788,6 @@ version = "0.8.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e5ce46fe64a9d73be07dcbe690a38ce1b293be448fd8ce1e6c1b8062c9f72c6a"
[[package]]
name = "neon-shmem"
version = "0.1.0"
dependencies = [
"nix 0.30.1",
"tempfile",
"thiserror 1.0.69",
"workspace_hack",
]
[[package]]
name = "never-say-never"
version = "6.6.666"
@@ -3837,13 +3821,12 @@ dependencies = [
[[package]]
name = "nix"
version = "0.30.1"
version = "0.27.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "74523f3a35e05aba87a1d978330aef40f67b0304ac79c1c00b294c9830543db6"
checksum = "2eb04e9c688eff1c89d72b407f168cf79bb9e867a9d3323ed6c01519eb9cc053"
dependencies = [
"bitflags 2.8.0",
"cfg-if",
"cfg_aliases",
"libc",
"memoffset 0.9.0",
]
@@ -3898,16 +3881,6 @@ dependencies = [
"winapi",
]
[[package]]
name = "nu-ansi-term"
version = "0.46.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "77a8165726e8236064dbb45459242600304b42a5ea24ee2948e18e023bf7ba84"
dependencies = [
"overload",
"winapi",
]
[[package]]
name = "num"
version = "0.4.1"
@@ -4112,7 +4085,7 @@ dependencies = [
"opentelemetry-http",
"opentelemetry-proto",
"opentelemetry_sdk",
"prost 0.13.5",
"prost 0.13.3",
"reqwest",
"thiserror 1.0.69",
]
@@ -4125,8 +4098,8 @@ checksum = "a6e05acbfada5ec79023c85368af14abd0b307c015e9064d249b2a950ef459a6"
dependencies = [
"opentelemetry",
"opentelemetry_sdk",
"prost 0.13.5",
"tonic 0.12.3",
"prost 0.13.3",
"tonic",
]
[[package]]
@@ -4192,12 +4165,6 @@ version = "0.5.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4030760ffd992bef45b0ae3f10ce1aba99e33464c90d14dd7c039884963ddc7a"
[[package]]
name = "overload"
version = "0.1.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b15813163c1d831bf4a13c3610c05c0d03b39feb07f7e09fa234dac9b15aaf39"
[[package]]
name = "p256"
version = "0.11.1"
@@ -4302,7 +4269,6 @@ dependencies = [
"enumset",
"fail",
"futures",
"hashlink",
"hex",
"hex-literal",
"http-utils",
@@ -4314,14 +4280,13 @@ dependencies = [
"jsonwebtoken",
"md5",
"metrics",
"nix 0.30.1",
"nix 0.27.1",
"num-traits",
"num_cpus",
"once_cell",
"pageserver_api",
"pageserver_client",
"pageserver_compaction",
"pageserver_page_api",
"pem",
"pin-project-lite",
"postgres-protocol",
@@ -4330,7 +4295,6 @@ dependencies = [
"postgres_connection",
"postgres_ffi",
"postgres_initdb",
"posthog_client_lite",
"pprof",
"pq_proto",
"procfs",
@@ -4341,7 +4305,7 @@ dependencies = [
"reqwest",
"rpds",
"rstest",
"rustls 0.23.27",
"rustls 0.23.18",
"scopeguard",
"send-future",
"serde",
@@ -4360,16 +4324,13 @@ dependencies = [
"tokio-epoll-uring",
"tokio-io-timeout",
"tokio-postgres",
"tokio-rustls 0.26.2",
"tokio-rustls 0.26.0",
"tokio-stream",
"tokio-tar",
"tokio-util",
"toml_edit",
"tonic 0.13.1",
"tonic-reflection",
"tracing",
"tracing-utils",
"twox-hash",
"url",
"utils",
"uuid",
@@ -4394,7 +4355,7 @@ dependencies = [
"humantime",
"humantime-serde",
"itertools 0.10.5",
"nix 0.30.1",
"nix 0.27.1",
"once_cell",
"postgres_backend",
"postgres_ffi",
@@ -4455,22 +4416,6 @@ dependencies = [
"workspace_hack",
]
[[package]]
name = "pageserver_page_api"
version = "0.1.0"
dependencies = [
"bytes",
"pageserver_api",
"postgres_ffi",
"prost 0.13.5",
"smallvec",
"thiserror 1.0.69",
"tonic 0.13.1",
"tonic-build",
"utils",
"workspace_hack",
]
[[package]]
name = "papaya"
version = "0.2.1"
@@ -4847,14 +4792,14 @@ dependencies = [
"bytes",
"once_cell",
"pq_proto",
"rustls 0.23.27",
"rustls 0.23.18",
"rustls-pemfile 2.1.1",
"serde",
"thiserror 1.0.69",
"tokio",
"tokio-postgres",
"tokio-postgres-rustls",
"tokio-rustls 0.26.2",
"tokio-rustls 0.26.0",
"tokio-util",
"tracing",
]
@@ -4908,16 +4853,11 @@ name = "posthog_client_lite"
version = "0.1.0"
dependencies = [
"anyhow",
"arc-swap",
"reqwest",
"serde",
"serde_json",
"sha2",
"thiserror 1.0.69",
"tokio",
"tokio-util",
"tracing",
"tracing-utils",
"workspace_hack",
]
@@ -4966,7 +4906,7 @@ dependencies = [
"inferno 0.12.0",
"num",
"paste",
"prost 0.13.5",
"prost 0.13.3",
]
[[package]]
@@ -5071,12 +5011,12 @@ dependencies = [
[[package]]
name = "prost"
version = "0.13.5"
version = "0.13.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "2796faa41db3ec313a31f7624d9286acf277b52de526150b7e69f3debf891ee5"
checksum = "7b0487d90e047de87f984913713b85c601c05609aad5b0df4b4573fbf69aa13f"
dependencies = [
"bytes",
"prost-derive 0.13.5",
"prost-derive 0.13.3",
]
[[package]]
@@ -5114,7 +5054,7 @@ dependencies = [
"once_cell",
"petgraph",
"prettyplease",
"prost 0.13.5",
"prost 0.13.3",
"prost-types 0.13.3",
"regex",
"syn 2.0.100",
@@ -5136,9 +5076,9 @@ dependencies = [
[[package]]
name = "prost-derive"
version = "0.13.5"
version = "0.13.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8a56d757972c98b346a9b766e3f02746cde6dd1cd1d1d563472929fdd74bec4d"
checksum = "e9552f850d5f0964a4e4d0bf306459ac29323ddfbae05e35a7c0d35cb0803cc5"
dependencies = [
"anyhow",
"itertools 0.12.1",
@@ -5162,7 +5102,7 @@ version = "0.13.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4759aa0d3a6232fb8dbdb97b61de2c20047c68aca932c7ed76da9d788508d670"
dependencies = [
"prost 0.13.5",
"prost 0.13.3",
]
[[package]]
@@ -5210,7 +5150,7 @@ dependencies = [
"hyper 0.14.30",
"hyper 1.4.1",
"hyper-util",
"indexmap 2.9.0",
"indexmap 2.0.1",
"ipnet",
"itertools 0.10.5",
"itoa",
@@ -5244,7 +5184,7 @@ dependencies = [
"rsa",
"rstest",
"rustc-hash 1.1.0",
"rustls 0.23.27",
"rustls 0.23.18",
"rustls-native-certs 0.8.0",
"rustls-pemfile 2.1.1",
"scopeguard",
@@ -5263,14 +5203,13 @@ dependencies = [
"tokio",
"tokio-postgres",
"tokio-postgres2",
"tokio-rustls 0.26.2",
"tokio-rustls 0.26.0",
"tokio-tungstenite 0.21.0",
"tokio-util",
"tracing",
"tracing-log",
"tracing-opentelemetry",
"tracing-subscriber",
"tracing-test",
"tracing-utils",
"try-lock",
"typed-json",
@@ -5487,13 +5426,13 @@ dependencies = [
"num-bigint",
"percent-encoding",
"pin-project-lite",
"rustls 0.23.27",
"rustls 0.23.18",
"rustls-native-certs 0.8.0",
"ryu",
"sha1_smol",
"socket2",
"tokio",
"tokio-rustls 0.26.2",
"tokio-rustls 0.26.0",
"tokio-util",
"url",
]
@@ -5941,15 +5880,15 @@ dependencies = [
[[package]]
name = "rustls"
version = "0.23.27"
version = "0.23.18"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "730944ca083c1c233a75c09f199e973ca499344a2b7ba9e755c457e86fb4a321"
checksum = "9c9cc1d47e243d655ace55ed38201c19ae02c148ae56412ab8750e8f0166ab7f"
dependencies = [
"log",
"once_cell",
"ring",
"rustls-pki-types",
"rustls-webpki 0.103.3",
"rustls-webpki 0.102.8",
"subtle",
"zeroize",
]
@@ -6038,17 +5977,6 @@ dependencies = [
"untrusted",
]
[[package]]
name = "rustls-webpki"
version = "0.103.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e4a72fe2bcf7a6ac6fd7d0b9e5cb68aeb7d4c0a0271730218b3e92d43b4eb435"
dependencies = [
"ring",
"rustls-pki-types",
"untrusted",
]
[[package]]
name = "rustversion"
version = "1.0.12"
@@ -6100,7 +6028,7 @@ dependencies = [
"regex",
"remote_storage",
"reqwest",
"rustls 0.23.27",
"rustls 0.23.18",
"safekeeper_api",
"safekeeper_client",
"scopeguard",
@@ -6117,7 +6045,7 @@ dependencies = [
"tokio",
"tokio-io-timeout",
"tokio-postgres",
"tokio-rustls 0.26.2",
"tokio-rustls 0.26.0",
"tokio-stream",
"tokio-tar",
"tokio-util",
@@ -6289,7 +6217,7 @@ checksum = "255914a8e53822abd946e2ce8baa41d4cded6b8e938913b7f7b9da5b7ab44335"
dependencies = [
"httpdate",
"reqwest",
"rustls 0.23.27",
"rustls 0.23.18",
"sentry-backtrace",
"sentry-contexts",
"sentry-core",
@@ -6718,11 +6646,11 @@ dependencies = [
"metrics",
"once_cell",
"parking_lot 0.12.1",
"prost 0.13.5",
"rustls 0.23.27",
"prost 0.13.3",
"rustls 0.23.18",
"tokio",
"tokio-rustls 0.26.2",
"tonic 0.13.1",
"tokio-rustls 0.26.0",
"tonic",
"tonic-build",
"tracing",
"utils",
@@ -6764,7 +6692,7 @@ dependencies = [
"regex",
"reqwest",
"routerify",
"rustls 0.23.27",
"rustls 0.23.18",
"rustls-native-certs 0.8.0",
"safekeeper_api",
"safekeeper_client",
@@ -6779,7 +6707,7 @@ dependencies = [
"tokio",
"tokio-postgres",
"tokio-postgres-rustls",
"tokio-rustls 0.26.2",
"tokio-rustls 0.26.0",
"tokio-util",
"tracing",
"utils",
@@ -6817,7 +6745,7 @@ dependencies = [
"postgres_ffi",
"remote_storage",
"reqwest",
"rustls 0.23.27",
"rustls 0.23.18",
"rustls-native-certs 0.8.0",
"serde",
"serde_json",
@@ -7351,10 +7279,10 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "04fb792ccd6bbcd4bba408eb8a292f70fc4a3589e5d793626f45190e6454b6ab"
dependencies = [
"ring",
"rustls 0.23.27",
"rustls 0.23.18",
"tokio",
"tokio-postgres",
"tokio-rustls 0.26.2",
"tokio-rustls 0.26.0",
"x509-certificate",
]
@@ -7398,11 +7326,12 @@ dependencies = [
[[package]]
name = "tokio-rustls"
version = "0.26.2"
version = "0.26.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8e727b36a1a0e8b74c376ac2211e40c2c8af09fb4013c60d910495810f008e9b"
checksum = "0c7bc40d0e5a97695bb96e27995cd3a08538541b0a846f65bba7a359f36700d4"
dependencies = [
"rustls 0.23.27",
"rustls 0.23.18",
"rustls-pki-types",
"tokio",
]
@@ -7500,7 +7429,7 @@ version = "0.22.14"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f21c7aaf97f1bd9ca9d4f9e73b0a6c74bd5afef56f2bc931943a6e1c37e04e38"
dependencies = [
"indexmap 2.9.0",
"indexmap 2.0.1",
"serde",
"serde_spanned",
"toml_datetime",
@@ -7519,41 +7448,18 @@ dependencies = [
"http 1.1.0",
"http-body 1.0.0",
"http-body-util",
"percent-encoding",
"pin-project",
"prost 0.13.5",
"tokio-stream",
"tower-layer",
"tower-service",
"tracing",
]
[[package]]
name = "tonic"
version = "0.13.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7e581ba15a835f4d9ea06c55ab1bd4dce26fc53752c69a04aac00703bfb49ba9"
dependencies = [
"async-trait",
"axum",
"base64 0.22.1",
"bytes",
"h2 0.4.4",
"http 1.1.0",
"http-body 1.0.0",
"http-body-util",
"hyper 1.4.1",
"hyper-timeout",
"hyper-util",
"percent-encoding",
"pin-project",
"prost 0.13.5",
"prost 0.13.3",
"rustls-native-certs 0.8.0",
"socket2",
"rustls-pemfile 2.1.1",
"tokio",
"tokio-rustls 0.26.2",
"tokio-rustls 0.26.0",
"tokio-stream",
"tower 0.5.2",
"tower 0.4.13",
"tower-layer",
"tower-service",
"tracing",
@@ -7561,9 +7467,9 @@ dependencies = [
[[package]]
name = "tonic-build"
version = "0.13.1"
version = "0.12.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "eac6f67be712d12f0b41328db3137e0d0757645d8904b4cb7d51cd9c2279e847"
checksum = "9557ce109ea773b399c9b9e5dca39294110b74f1f342cb347a80d1fce8c26a11"
dependencies = [
"prettyplease",
"proc-macro2",
@@ -7573,19 +7479,6 @@ dependencies = [
"syn 2.0.100",
]
[[package]]
name = "tonic-reflection"
version = "0.13.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f9687bd5bfeafebdded2356950f278bba8226f0b32109537c4253406e09aafe1"
dependencies = [
"prost 0.13.5",
"prost-types 0.13.3",
"tokio",
"tokio-stream",
"tonic 0.13.1",
]
[[package]]
name = "tower"
version = "0.4.13"
@@ -7594,11 +7487,16 @@ checksum = "b8fa9be0de6cf49e536ce1851f987bd21a43b771b09473c3549a6c853db37c1c"
dependencies = [
"futures-core",
"futures-util",
"indexmap 1.9.3",
"pin-project",
"pin-project-lite",
"rand 0.8.5",
"slab",
"tokio",
"tokio-util",
"tower-layer",
"tower-service",
"tracing",
]
[[package]]
@@ -7609,12 +7507,9 @@ checksum = "d039ad9159c98b70ecfd540b2573b97f7f52c3e8d9f8ad57a24b916a536975f9"
dependencies = [
"futures-core",
"futures-util",
"indexmap 2.9.0",
"pin-project-lite",
"slab",
"sync_wrapper 1.0.1",
"tokio",
"tokio-util",
"tower-layer",
"tower-service",
"tracing",
@@ -7765,7 +7660,6 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e8189decb5ac0fa7bc8b96b7cb9b2701d60d48805aca84a238004d665fcc4008"
dependencies = [
"matchers",
"nu-ansi-term",
"once_cell",
"regex",
"serde",
@@ -7779,27 +7673,6 @@ dependencies = [
"tracing-serde",
]
[[package]]
name = "tracing-test"
version = "0.2.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "557b891436fe0d5e0e363427fc7f217abf9ccd510d5136549847bdcbcd011d68"
dependencies = [
"tracing-core",
"tracing-subscriber",
"tracing-test-macro",
]
[[package]]
name = "tracing-test-macro"
version = "0.2.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "04659ddb06c87d233c566112c1c9c5b9e98256d9af50ec3bc9c8327f873a7568"
dependencies = [
"quote",
"syn 2.0.100",
]
[[package]]
name = "tracing-utils"
version = "0.1.0"
@@ -7942,7 +7815,7 @@ dependencies = [
"base64 0.22.1",
"log",
"once_cell",
"rustls 0.23.27",
"rustls 0.23.18",
"rustls-pki-types",
"url",
"webpki-roots",
@@ -8025,7 +7898,7 @@ dependencies = [
"humantime",
"jsonwebtoken",
"metrics",
"nix 0.30.1",
"nix 0.27.1",
"once_cell",
"pem",
"pin-project-lite",
@@ -8137,7 +8010,7 @@ dependencies = [
"pageserver_api",
"postgres_ffi",
"pprof",
"prost 0.13.5",
"prost 0.13.3",
"remote_storage",
"serde",
"serde_json",
@@ -8557,8 +8430,6 @@ dependencies = [
"ahash",
"anstream",
"anyhow",
"axum",
"axum-core",
"base64 0.13.1",
"base64 0.21.7",
"base64ct",
@@ -8581,8 +8452,10 @@ dependencies = [
"fail",
"form_urlencoded",
"futures-channel",
"futures-core",
"futures-executor",
"futures-io",
"futures-task",
"futures-util",
"generic-array",
"getrandom 0.2.11",
@@ -8593,14 +8466,14 @@ dependencies = [
"hyper 0.14.30",
"hyper 1.4.1",
"hyper-util",
"indexmap 2.9.0",
"indexmap 1.9.3",
"indexmap 2.0.1",
"itertools 0.12.1",
"lazy_static",
"libc",
"log",
"memchr",
"nix 0.26.4",
"nix 0.30.1",
"nom",
"num",
"num-bigint",
@@ -8612,18 +8485,19 @@ dependencies = [
"once_cell",
"p256 0.13.2",
"parquet",
"percent-encoding",
"prettyplease",
"proc-macro2",
"prost 0.13.5",
"prost 0.13.3",
"quote",
"rand 0.8.5",
"regex",
"regex-automata 0.4.3",
"regex-syntax 0.8.2",
"reqwest",
"rustls 0.23.27",
"rustls 0.23.18",
"rustls-pki-types",
"rustls-webpki 0.103.3",
"rustls-webpki 0.102.8",
"scopeguard",
"sec1 0.7.3",
"serde",
@@ -8641,15 +8515,15 @@ dependencies = [
"time",
"time-macros",
"tokio",
"tokio-rustls 0.26.2",
"tokio-rustls 0.26.0",
"tokio-stream",
"tokio-util",
"toml_edit",
"tower 0.5.2",
"tonic",
"tower 0.4.13",
"tracing",
"tracing-core",
"tracing-log",
"tracing-subscriber",
"url",
"uuid",
"zeroize",

View File

@@ -9,7 +9,6 @@ members = [
"pageserver/ctl",
"pageserver/client",
"pageserver/pagebench",
"pageserver/page_api",
"proxy",
"safekeeper",
"safekeeper/client",
@@ -24,7 +23,6 @@ members = [
"libs/postgres_ffi",
"libs/safekeeper_api",
"libs/desim",
"libs/neon-shmem",
"libs/utils",
"libs/consumption_metrics",
"libs/postgres_backend",
@@ -129,7 +127,7 @@ md5 = "0.7.0"
measured = { version = "0.0.22", features=["lasso"] }
measured-process = { version = "0.0.22" }
memoffset = "0.9"
nix = { version = "0.30.1", features = ["dir", "fs", "mman", "process", "socket", "signal", "poll"] }
nix = { version = "0.27", features = ["dir", "fs", "process", "socket", "signal", "poll"] }
# Do not update to >= 7.0.0, at least. The update will have a significant impact
# on compute startup metrics (start_postgres_ms), >= 25% degradation.
notify = "6.0.0"
@@ -149,7 +147,7 @@ pin-project-lite = "0.2"
pprof = { version = "0.14", features = ["criterion", "flamegraph", "frame-pointer", "prost-codec"] }
procfs = "0.16"
prometheus = {version = "0.13", default-features=false, features = ["process"]} # removes protobuf dependency
prost = "0.13.5"
prost = "0.13"
rand = "0.8"
redis = { version = "0.29.2", features = ["tokio-rustls-comp", "keep-alive"] }
regex = "1.10.2"
@@ -199,8 +197,7 @@ tokio-tar = "0.3"
tokio-util = { version = "0.7.10", features = ["io", "rt"] }
toml = "0.8"
toml_edit = "0.22"
tonic = { version = "0.13.1", default-features = false, features = ["channel", "codegen", "prost", "router", "server", "tls-ring", "tls-native-roots"] }
tonic-reflection = { version = "0.13.1", features = ["server"] }
tonic = {version = "0.12.3", default-features = false, features = ["channel", "tls", "tls-roots"]}
tower = { version = "0.5.2", default-features = false }
tower-http = { version = "0.6.2", features = ["auth", "request-id", "trace"] }
@@ -247,7 +244,6 @@ azure_storage_blobs = { git = "https://github.com/neondatabase/azure-sdk-for-rus
## Local libraries
compute_api = { version = "0.1", path = "./libs/compute_api/" }
consumption_metrics = { version = "0.1", path = "./libs/consumption_metrics/" }
desim = { version = "0.1", path = "./libs/desim" }
endpoint_storage = { version = "0.0.1", path = "./endpoint_storage/" }
http-utils = { version = "0.1", path = "./libs/http-utils/" }
metrics = { version = "0.1", path = "./libs/metrics/" }
@@ -255,24 +251,23 @@ pageserver = { path = "./pageserver" }
pageserver_api = { version = "0.1", path = "./libs/pageserver_api/" }
pageserver_client = { path = "./pageserver/client" }
pageserver_compaction = { version = "0.1", path = "./pageserver/compaction/" }
pageserver_page_api = { path = "./pageserver/page_api" }
postgres_backend = { version = "0.1", path = "./libs/postgres_backend/" }
postgres_connection = { version = "0.1", path = "./libs/postgres_connection/" }
postgres_ffi = { version = "0.1", path = "./libs/postgres_ffi/" }
postgres_initdb = { path = "./libs/postgres_initdb" }
posthog_client_lite = { version = "0.1", path = "./libs/posthog_client_lite" }
pq_proto = { version = "0.1", path = "./libs/pq_proto/" }
remote_storage = { version = "0.1", path = "./libs/remote_storage/" }
safekeeper_api = { version = "0.1", path = "./libs/safekeeper_api" }
safekeeper_client = { path = "./safekeeper/client" }
desim = { version = "0.1", path = "./libs/desim" }
storage_broker = { version = "0.1", path = "./storage_broker/" } # Note: main broker code is inside the binary crate, so linking with the library shouldn't be heavy.
storage_controller_client = { path = "./storage_controller/client" }
tenant_size_model = { version = "0.1", path = "./libs/tenant_size_model/" }
tracing-utils = { version = "0.1", path = "./libs/tracing-utils/" }
utils = { version = "0.1", path = "./libs/utils/" }
vm_monitor = { version = "0.1", path = "./libs/vm_monitor/" }
wal_decoder = { version = "0.1", path = "./libs/wal_decoder" }
walproposer = { version = "0.1", path = "./libs/walproposer/" }
wal_decoder = { version = "0.1", path = "./libs/wal_decoder" }
## Common library dependency
workspace_hack = { version = "0.1", path = "./workspace_hack/" }
@@ -282,7 +277,7 @@ criterion = "0.5.1"
rcgen = "0.13"
rstest = "0.18"
camino-tempfile = "1.0.2"
tonic-build = "0.13.1"
tonic-build = "0.12"
[patch.crates-io]

View File

@@ -155,7 +155,7 @@ RUN set -e \
# Keep the version the same as in compute/compute-node.Dockerfile and
# test_runner/regress/test_compute_metrics.py.
ENV SQL_EXPORTER_VERSION=0.17.3
ENV SQL_EXPORTER_VERSION=0.17.0
RUN curl -fsSL \
"https://github.com/burningalchemist/sql_exporter/releases/download/${SQL_EXPORTER_VERSION}/sql_exporter-${SQL_EXPORTER_VERSION}.linux-$(case "$(uname -m)" in x86_64) echo amd64;; aarch64) echo arm64;; esac).tar.gz" \
--output sql_exporter.tar.gz \
@@ -292,7 +292,7 @@ WORKDIR /home/nonroot
# Rust
# Please keep the version of llvm (installed above) in sync with rust llvm (`rustc --version --verbose | grep LLVM`)
ENV RUSTC_VERSION=1.87.0
ENV RUSTC_VERSION=1.86.0
ENV RUSTUP_HOME="/home/nonroot/.rustup"
ENV PATH="/home/nonroot/.cargo/bin:${PATH}"
ARG RUSTFILT_VERSION=0.2.1
@@ -310,13 +310,13 @@ RUN curl -sSO https://static.rust-lang.org/rustup/dist/$(uname -m)-unknown-linux
. "$HOME/.cargo/env" && \
cargo --version && rustup --version && \
rustup component add llvm-tools rustfmt clippy && \
cargo install rustfilt --version ${RUSTFILT_VERSION} --locked && \
cargo install cargo-hakari --version ${CARGO_HAKARI_VERSION} --locked && \
cargo install cargo-deny --version ${CARGO_DENY_VERSION} --locked && \
cargo install cargo-hack --version ${CARGO_HACK_VERSION} --locked && \
cargo install cargo-nextest --version ${CARGO_NEXTEST_VERSION} --locked && \
cargo install cargo-chef --version ${CARGO_CHEF_VERSION} --locked && \
cargo install diesel_cli --version ${CARGO_DIESEL_CLI_VERSION} --locked \
cargo install rustfilt --version ${RUSTFILT_VERSION} && \
cargo install cargo-hakari --version ${CARGO_HAKARI_VERSION} && \
cargo install cargo-deny --locked --version ${CARGO_DENY_VERSION} && \
cargo install cargo-hack --version ${CARGO_HACK_VERSION} && \
cargo install cargo-nextest --version ${CARGO_NEXTEST_VERSION} && \
cargo install cargo-chef --locked --version ${CARGO_CHEF_VERSION} && \
cargo install diesel_cli --version ${CARGO_DIESEL_CLI_VERSION} \
--features postgres-bundled --no-default-features && \
rm -rf /home/nonroot/.cargo/registry && \
rm -rf /home/nonroot/.cargo/git

View File

@@ -582,38 +582,6 @@ RUN make -j $(getconf _NPROCESSORS_ONLN) && \
make -j $(getconf _NPROCESSORS_ONLN) install && \
echo 'trusted = true' >> /usr/local/pgsql/share/extension/hypopg.control
#########################################################################################
#
# Layer "online_advisor-build"
# compile online_advisor extension
#
#########################################################################################
FROM build-deps AS online_advisor-src
ARG PG_VERSION
# online_advisor supports all Postgres version starting from PG14, but prior to PG17 has to be included in preload_shared_libraries
# last release 1.0 - May 15, 2025
WORKDIR /ext-src
RUN case "${PG_VERSION:?}" in \
"v17") \
;; \
*) \
echo "skipping the version of online_advistor for $PG_VERSION" && exit 0 \
;; \
esac && \
wget https://github.com/knizhnik/online_advisor/archive/refs/tags/1.0.tar.gz -O online_advisor.tar.gz && \
echo "059b7d9e5a90013a58bdd22e9505b88406ce05790675eb2d8434e5b215652d54 online_advisor.tar.gz" | sha256sum --check && \
mkdir online_advisor-src && cd online_advisor-src && tar xzf ../online_advisor.tar.gz --strip-components=1 -C .
FROM pg-build AS online_advisor-build
COPY --from=online_advisor-src /ext-src/ /ext-src/
WORKDIR /ext-src/
RUN if [ -d online_advisor-src ]; then \
cd online_advisor-src && \
make -j install && \
echo 'trusted = true' >> /usr/local/pgsql/share/extension/online_advisor.control; \
fi
#########################################################################################
#
# Layer "pg_hashids-build"
@@ -1680,7 +1648,6 @@ COPY --from=pg_jsonschema-build /usr/local/pgsql/ /usr/local/pgsql/
COPY --from=pg_graphql-build /usr/local/pgsql/ /usr/local/pgsql/
COPY --from=pg_tiktoken-build /usr/local/pgsql/ /usr/local/pgsql/
COPY --from=hypopg-build /usr/local/pgsql/ /usr/local/pgsql/
COPY --from=online_advisor-build /usr/local/pgsql/ /usr/local/pgsql/
COPY --from=pg_hashids-build /usr/local/pgsql/ /usr/local/pgsql/
COPY --from=rum-build /usr/local/pgsql/ /usr/local/pgsql/
COPY --from=pgtap-build /usr/local/pgsql/ /usr/local/pgsql/
@@ -1784,17 +1751,17 @@ ARG TARGETARCH
RUN if [ "$TARGETARCH" = "amd64" ]; then\
postgres_exporter_sha256='59aa4a7bb0f7d361f5e05732f5ed8c03cc08f78449cef5856eadec33a627694b';\
pgbouncer_exporter_sha256='c9f7cf8dcff44f0472057e9bf52613d93f3ffbc381ad7547a959daa63c5e84ac';\
sql_exporter_sha256='9a41127a493e8bfebfe692bf78c7ed2872a58a3f961ee534d1b0da9ae584aaab';\
sql_exporter_sha256='38e439732bbf6e28ca4a94d7bc3686d3fa1abdb0050773d5617a9efdb9e64d08';\
else\
postgres_exporter_sha256='d1dedea97f56c6d965837bfd1fbb3e35a3b4a4556f8cccee8bd513d8ee086124';\
pgbouncer_exporter_sha256='217c4afd7e6492ae904055bc14fe603552cf9bac458c063407e991d68c519da3';\
sql_exporter_sha256='530e6afc77c043497ed965532c4c9dfa873bc2a4f0b3047fad367715c0081d6a';\
sql_exporter_sha256='11918b00be6e2c3a67564adfdb2414fdcbb15a5db76ea17d1d1a944237a893c6';\
fi\
&& curl -sL https://github.com/prometheus-community/postgres_exporter/releases/download/v0.17.1/postgres_exporter-0.17.1.linux-${TARGETARCH}.tar.gz\
| tar xzf - --strip-components=1 -C.\
&& curl -sL https://github.com/prometheus-community/pgbouncer_exporter/releases/download/v0.10.2/pgbouncer_exporter-0.10.2.linux-${TARGETARCH}.tar.gz\
| tar xzf - --strip-components=1 -C.\
&& curl -sL https://github.com/burningalchemist/sql_exporter/releases/download/0.17.3/sql_exporter-0.17.3.linux-${TARGETARCH}.tar.gz\
&& curl -sL https://github.com/burningalchemist/sql_exporter/releases/download/0.17.0/sql_exporter-0.17.0.linux-${TARGETARCH}.tar.gz\
| tar xzf - --strip-components=1 -C.\
&& echo "${postgres_exporter_sha256} postgres_exporter" | sha256sum -c -\
&& echo "${pgbouncer_exporter_sha256} pgbouncer_exporter" | sha256sum -c -\
@@ -1847,7 +1814,7 @@ COPY docker-compose/ext-src/ /ext-src/
COPY --from=pg-build /postgres /postgres
#COPY --from=postgis-src /ext-src/ /ext-src/
COPY --from=plv8-src /ext-src/ /ext-src/
COPY --from=h3-pg-src /ext-src/h3-pg-src /ext-src/h3-pg-src
#COPY --from=h3-pg-src /ext-src/ /ext-src/
COPY --from=postgresql-unit-src /ext-src/ /ext-src/
COPY --from=pgvector-src /ext-src/ /ext-src/
COPY --from=pgjwt-src /ext-src/ /ext-src/
@@ -1856,7 +1823,6 @@ COPY --from=pgjwt-src /ext-src/ /ext-src/
COPY --from=pg_graphql-src /ext-src/ /ext-src/
#COPY --from=pg_tiktoken-src /ext-src/ /ext-src/
COPY --from=hypopg-src /ext-src/ /ext-src/
COPY --from=online_advisor-src /ext-src/ /ext-src/
COPY --from=pg_hashids-src /ext-src/ /ext-src/
COPY --from=rum-src /ext-src/ /ext-src/
COPY --from=pgtap-src /ext-src/ /ext-src/

View File

@@ -7,7 +7,7 @@ index 255e616..1c6edb7 100644
RelationGetRelationName(index));
+#ifdef NEON_SMGR
+ smgr_start_unlogged_build(RelationGetSmgr(index));
+ smgr_start_unlogged_build(index->rd_smgr);
+#endif
+
initRumState(&buildstate.rumstate, index);
@@ -18,7 +18,7 @@ index 255e616..1c6edb7 100644
rumUpdateStats(index, &buildstate.buildStats, buildstate.rumstate.isBuild);
+#ifdef NEON_SMGR
+ smgr_finish_unlogged_build_phase_1(RelationGetSmgr(index));
+ smgr_finish_unlogged_build_phase_1(index->rd_smgr);
+#endif
+
/*
@@ -29,7 +29,7 @@ index 255e616..1c6edb7 100644
}
+#ifdef NEON_SMGR
+ smgr_end_unlogged_build(RelationGetSmgr(index));
+ smgr_end_unlogged_build(index->rd_smgr);
+#endif
+
/*

View File

@@ -136,10 +136,6 @@ struct Cli {
requires = "compute-id"
)]
pub control_plane_uri: Option<String>,
/// Interval in seconds for collecting installed extensions statistics
#[arg(long, default_value = "3600")]
pub installed_extensions_collection_interval: u64,
}
fn main() -> Result<()> {
@@ -183,7 +179,6 @@ fn main() -> Result<()> {
cgroup: cli.cgroup,
#[cfg(target_os = "linux")]
vm_monitor_addr: cli.vm_monitor_addr,
installed_extensions_collection_interval: cli.installed_extensions_collection_interval,
},
config,
)?;

View File

@@ -339,8 +339,6 @@ async fn run_dump_restore(
destination_connstring: String,
) -> Result<(), anyhow::Error> {
let dumpdir = workdir.join("dumpdir");
let num_jobs = num_cpus::get().to_string();
info!("using {num_jobs} jobs for dump/restore");
let common_args = [
// schema mapping (prob suffices to specify them on one side)
@@ -356,7 +354,7 @@ async fn run_dump_restore(
"directory".to_string(),
// concurrency
"--jobs".to_string(),
num_jobs,
num_cpus::get().to_string(),
// progress updates
"--verbose".to_string(),
];

View File

@@ -97,9 +97,6 @@ pub struct ComputeNodeParams {
/// the address of extension storage proxy gateway
pub remote_ext_base_url: Option<String>,
/// Interval for installed extensions collection
pub installed_extensions_collection_interval: u64,
}
/// Compute node info shared across several `compute_ctl` threads.
@@ -698,18 +695,25 @@ impl ComputeNode {
let log_directory_path = Path::new(&self.params.pgdata).join("log");
let log_directory_path = log_directory_path.to_string_lossy().to_string();
// Add project_id,endpoint_id to identify the logs.
// Add project_id,endpoint_id tag to identify the logs.
//
// These ids are passed from cplane,
let endpoint_id = pspec.spec.endpoint_id.as_deref().unwrap_or("");
let project_id = pspec.spec.project_id.as_deref().unwrap_or("");
// for backwards compatibility (old computes that don't have them),
// we set them to None.
// TODO: Clean up this code when all computes have them.
let tag: Option<String> = match (
pspec.spec.project_id.as_deref(),
pspec.spec.endpoint_id.as_deref(),
) {
(Some(project_id), Some(endpoint_id)) => {
Some(format!("{project_id}/{endpoint_id}"))
}
(Some(project_id), None) => Some(format!("{project_id}/None")),
(None, Some(endpoint_id)) => Some(format!("None,{endpoint_id}")),
(None, None) => None,
};
configure_audit_rsyslog(
log_directory_path.clone(),
endpoint_id,
project_id,
&remote_endpoint,
)?;
configure_audit_rsyslog(log_directory_path.clone(), tag, &remote_endpoint)?;
// Launch a background task to clean up the audit logs
launch_pgaudit_gc(log_directory_path);
@@ -745,7 +749,17 @@ impl ComputeNode {
let conf = self.get_tokio_conn_conf(None);
tokio::task::spawn(async {
let _ = installed_extensions(conf).await;
let res = get_installed_extensions(conf).await;
match res {
Ok(extensions) => {
info!(
"[NEON_EXT_STAT] {}",
serde_json::to_string(&extensions)
.expect("failed to serialize extensions list")
);
}
Err(err) => error!("could not get installed extensions: {err:?}"),
}
});
}
@@ -775,9 +789,6 @@ impl ComputeNode {
// Log metrics so that we can search for slow operations in logs
info!(?metrics, postmaster_pid = %postmaster_pid, "compute start finished");
// Spawn the extension stats background task
self.spawn_extension_stats_task();
if pspec.spec.prewarm_lfc_on_startup {
self.prewarm_lfc();
}
@@ -2188,41 +2199,6 @@ LIMIT 100",
info!("Pageserver config changed");
}
}
pub fn spawn_extension_stats_task(&self) {
let conf = self.tokio_conn_conf.clone();
let installed_extensions_collection_interval =
self.params.installed_extensions_collection_interval;
tokio::spawn(async move {
// An initial sleep is added to ensure that two collections don't happen at the same time.
// The first collection happens during compute startup.
tokio::time::sleep(tokio::time::Duration::from_secs(
installed_extensions_collection_interval,
))
.await;
let mut interval = tokio::time::interval(tokio::time::Duration::from_secs(
installed_extensions_collection_interval,
));
loop {
interval.tick().await;
let _ = installed_extensions(conf.clone()).await;
}
});
}
}
pub async fn installed_extensions(conf: tokio_postgres::Config) -> Result<()> {
let res = get_installed_extensions(conf).await;
match res {
Ok(extensions) => {
info!(
"[NEON_EXT_STAT] {}",
serde_json::to_string(&extensions).expect("failed to serialize extensions list")
);
}
Err(err) => error!("could not get installed extensions: {err:?}"),
}
Ok(())
}
pub fn forward_termination_signal() {

View File

@@ -2,24 +2,10 @@
module(load="imfile")
# Input configuration for log files in the specified directory
# The messages can be multiline. The start of the message is a timestamp
# in "%Y-%m-%d %H:%M:%S.%3N GMT" (so timezone hardcoded).
# Replace log_directory with the directory containing the log files
input(type="imfile" File="{log_directory}/*.log"
Tag="pgaudit_log" Severity="info" Facility="local5"
startmsg.regex="^[[:digit:]]{{4}}-[[:digit:]]{{2}}-[[:digit:]]{{2}} [[:digit:]]{{2}}:[[:digit:]]{{2}}:[[:digit:]]{{2}}.[[:digit:]]{{3}} GMT,")
# Replace {log_directory} with the directory containing the log files
input(type="imfile" File="{log_directory}/*.log" Tag="{tag}" Severity="info" Facility="local0")
# the directory to store rsyslog state files
global(workDirectory="/var/log/rsyslog")
# Construct json, endpoint_id and project_id as additional metadata
set $.json_log!endpoint_id = "{endpoint_id}";
set $.json_log!project_id = "{project_id}";
set $.json_log!msg = $msg;
# Template suitable for rfc5424 syslog format
template(name="PgAuditLog" type="string"
string="<%PRI%>1 %TIMESTAMP:::date-rfc3339% %HOSTNAME% - - - - %$.json_log%")
# Forward to remote syslog receiver (@@<hostname>:<port>;format
local5.info @@{remote_endpoint};PgAuditLog
# Forward logs to remote syslog server
*.* @@{remote_endpoint}

View File

@@ -213,10 +213,8 @@ impl Escaping for PgIdent {
// Find the first suitable tag that is not present in the string.
// Postgres' max role/DB name length is 63 bytes, so even in the
// worst case it won't take long. Outer tag is always `tag + "x"`,
// so if `tag` is not present in the string, `outer_tag` is not
// present in the string either.
while self.contains(&tag.to_string()) {
// worst case it won't take long.
while self.contains(&format!("${tag}$")) || self.contains(&format!("${outer_tag}$")) {
tag += "x";
outer_tag = tag.clone() + "x";
}

View File

@@ -27,40 +27,6 @@ fn get_rsyslog_pid() -> Option<String> {
}
}
fn wait_for_rsyslog_pid() -> Result<String, anyhow::Error> {
const MAX_WAIT: Duration = Duration::from_secs(5);
const INITIAL_SLEEP: Duration = Duration::from_millis(2);
let mut sleep_duration = INITIAL_SLEEP;
let start = std::time::Instant::now();
let mut attempts = 1;
for attempt in 1.. {
attempts = attempt;
match get_rsyslog_pid() {
Some(pid) => return Ok(pid),
None => {
if start.elapsed() >= MAX_WAIT {
break;
}
info!(
"rsyslogd is not running, attempt {}. Sleeping for {} ms",
attempt,
sleep_duration.as_millis()
);
std::thread::sleep(sleep_duration);
sleep_duration *= 2;
}
}
}
Err(anyhow::anyhow!(
"rsyslogd is not running after waiting for {} seconds and {} attempts",
attempts,
start.elapsed().as_secs()
))
}
// Restart rsyslogd to apply the new configuration.
// This is necessary, because there is no other way to reload the rsyslog configuration.
//
@@ -70,29 +36,27 @@ fn wait_for_rsyslog_pid() -> Result<String, anyhow::Error> {
// TODO: test it properly
//
fn restart_rsyslog() -> Result<()> {
let old_pid = get_rsyslog_pid().context("rsyslogd is not running")?;
info!("rsyslogd is running with pid: {}, restart it", old_pid);
// kill it to restart
let _ = Command::new("pkill")
.arg("rsyslogd")
.output()
.context("Failed to restart rsyslogd")?;
// ensure rsyslogd is running
wait_for_rsyslog_pid()?;
.context("Failed to stop rsyslogd")?;
Ok(())
}
pub fn configure_audit_rsyslog(
log_directory: String,
endpoint_id: &str,
project_id: &str,
tag: Option<String>,
remote_endpoint: &str,
) -> Result<()> {
let config_content: String = format!(
include_str!("config_template/compute_audit_rsyslog_template.conf"),
log_directory = log_directory,
endpoint_id = endpoint_id,
project_id = project_id,
tag = tag.unwrap_or("".to_string()),
remote_endpoint = remote_endpoint
);
@@ -167,11 +131,15 @@ pub fn configure_postgres_logs_export(conf: PostgresLogsRsyslogConfig) -> Result
return Ok(());
}
// Nothing to configure
// When new config is empty we can simply remove the configuration file.
if new_config.is_empty() {
// When the configuration is removed, PostgreSQL will stop sending data
// to the files watched by rsyslog, so restarting rsyslog is more effort
// than just ignoring this change.
info!("removing rsyslog config file: {}", POSTGRES_LOGS_CONF_PATH);
match std::fs::remove_file(POSTGRES_LOGS_CONF_PATH) {
Ok(_) => {}
Err(err) if err.kind() == ErrorKind::NotFound => {}
Err(err) => return Err(err.into()),
}
restart_rsyslog()?;
return Ok(());
}

View File

@@ -71,14 +71,6 @@ test.escaping = 'here''s a backslash \\ and a quote '' and a double-quote " hoor
("name$$$", ("$x$name$$$$x$", "xx")),
("name$$$$", ("$x$name$$$$$x$", "xx")),
("name$x$", ("$xx$name$x$$xx$", "xxx")),
("x", ("$xx$x$xx$", "xxx")),
("xx", ("$xxx$xx$xxx$", "xxxx")),
("$x", ("$xx$$x$xx$", "xxx")),
("x$", ("$xx$x$$xx$", "xxx")),
("$x$", ("$xx$$x$$xx$", "xxx")),
("xx$", ("$xxx$xx$$xxx$", "xxxx")),
("$xx", ("$xxx$$xx$xxx$", "xxxx")),
("$xx$", ("$xxx$$xx$$xxx$", "xxxx")),
];
for (input, expected) in test_cases {

View File

@@ -2,10 +2,8 @@
[pageserver]
listen_pg_addr = '127.0.0.1:64000'
listen_http_addr = '127.0.0.1:9898'
listen_grpc_addr = '127.0.0.1:51051'
pg_auth_type = 'Trust'
http_auth_type = 'Trust'
grpc_auth_type = 'Trust'
[[safekeepers]]
id = 1

View File

@@ -4,10 +4,8 @@
id=1
listen_pg_addr = '127.0.0.1:64000'
listen_http_addr = '127.0.0.1:9898'
listen_grpc_addr = '127.0.0.1:51051'
pg_auth_type = 'Trust'
http_auth_type = 'Trust'
grpc_auth_type = 'Trust'
[[safekeepers]]
id = 1

View File

@@ -14,7 +14,7 @@
use std::ffi::OsStr;
use std::io::Write;
use std::os::fd::AsFd;
use std::os::unix::prelude::AsRawFd;
use std::os::unix::process::CommandExt;
use std::path::Path;
use std::process::Command;
@@ -356,7 +356,7 @@ where
let file = pid_file::claim_for_current_process(&path).expect("claim pid file");
// Remove the FD_CLOEXEC flag on the pidfile descriptor so that the pidfile
// remains locked after exec.
nix::fcntl::fcntl(file.as_fd(), FcntlArg::F_SETFD(FdFlag::empty()))
nix::fcntl::fcntl(file.as_raw_fd(), FcntlArg::F_SETFD(FdFlag::empty()))
.expect("remove FD_CLOEXEC");
// Don't run drop(file), it would close the file before we actually exec.
std::mem::forget(file);

View File

@@ -8,6 +8,7 @@
use std::borrow::Cow;
use std::collections::{BTreeSet, HashMap};
use std::fs::File;
use std::os::fd::AsRawFd;
use std::path::PathBuf;
use std::process::exit;
use std::str::FromStr;
@@ -30,9 +31,8 @@ use control_plane::safekeeper::SafekeeperNode;
use control_plane::storage_controller::{
NeonStorageControllerStartArgs, NeonStorageControllerStopArgs, StorageController,
};
use nix::fcntl::{Flock, FlockArg};
use nix::fcntl::{FlockArg, flock};
use pageserver_api::config::{
DEFAULT_GRPC_LISTEN_PORT as DEFAULT_PAGESERVER_GRPC_PORT,
DEFAULT_HTTP_LISTEN_PORT as DEFAULT_PAGESERVER_HTTP_PORT,
DEFAULT_PG_LISTEN_PORT as DEFAULT_PAGESERVER_PG_PORT,
};
@@ -749,16 +749,16 @@ struct TimelineTreeEl {
/// A flock-based guard over the neon_local repository directory
struct RepoLock {
_file: Flock<File>,
_file: File,
}
impl RepoLock {
fn new() -> Result<Self> {
let repo_dir = File::open(local_env::base_path())?;
match Flock::lock(repo_dir, FlockArg::LockExclusive) {
Ok(f) => Ok(Self { _file: f }),
Err((_, e)) => Err(e).context("flock error"),
}
let repo_dir_fd = repo_dir.as_raw_fd();
flock(repo_dir_fd, FlockArg::LockExclusive)?;
Ok(Self { _file: repo_dir })
}
}
@@ -1008,16 +1008,13 @@ fn handle_init(args: &InitCmdArgs) -> anyhow::Result<LocalEnv> {
let pageserver_id = NodeId(DEFAULT_PAGESERVER_ID.0 + i as u64);
let pg_port = DEFAULT_PAGESERVER_PG_PORT + i;
let http_port = DEFAULT_PAGESERVER_HTTP_PORT + i;
let grpc_port = DEFAULT_PAGESERVER_GRPC_PORT + i;
NeonLocalInitPageserverConf {
id: pageserver_id,
listen_pg_addr: format!("127.0.0.1:{pg_port}"),
listen_http_addr: format!("127.0.0.1:{http_port}"),
listen_https_addr: None,
listen_grpc_addr: Some(format!("127.0.0.1:{grpc_port}")),
pg_auth_type: AuthType::Trust,
http_auth_type: AuthType::Trust,
grpc_auth_type: AuthType::Trust,
other: Default::default(),
// Typical developer machines use disks with slow fsync, and we don't care
// about data integrity: disable disk syncs.
@@ -1279,7 +1276,6 @@ async fn handle_timeline(cmd: &TimelineCmd, env: &mut local_env::LocalEnv) -> Re
mode: pageserver_api::models::TimelineCreateRequestMode::Branch {
ancestor_timeline_id,
ancestor_start_lsn: start_lsn,
read_only: false,
pg_version: None,
},
};

View File

@@ -278,10 +278,8 @@ pub struct PageServerConf {
pub listen_pg_addr: String,
pub listen_http_addr: String,
pub listen_https_addr: Option<String>,
pub listen_grpc_addr: Option<String>,
pub pg_auth_type: AuthType,
pub http_auth_type: AuthType,
pub grpc_auth_type: AuthType,
pub no_sync: bool,
}
@@ -292,10 +290,8 @@ impl Default for PageServerConf {
listen_pg_addr: String::new(),
listen_http_addr: String::new(),
listen_https_addr: None,
listen_grpc_addr: None,
pg_auth_type: AuthType::Trust,
http_auth_type: AuthType::Trust,
grpc_auth_type: AuthType::Trust,
no_sync: false,
}
}
@@ -310,10 +306,8 @@ pub struct NeonLocalInitPageserverConf {
pub listen_pg_addr: String,
pub listen_http_addr: String,
pub listen_https_addr: Option<String>,
pub listen_grpc_addr: Option<String>,
pub pg_auth_type: AuthType,
pub http_auth_type: AuthType,
pub grpc_auth_type: AuthType,
#[serde(default, skip_serializing_if = "std::ops::Not::not")]
pub no_sync: bool,
#[serde(flatten)]
@@ -327,10 +321,8 @@ impl From<&NeonLocalInitPageserverConf> for PageServerConf {
listen_pg_addr,
listen_http_addr,
listen_https_addr,
listen_grpc_addr,
pg_auth_type,
http_auth_type,
grpc_auth_type,
no_sync,
other: _,
} = conf;
@@ -339,9 +331,7 @@ impl From<&NeonLocalInitPageserverConf> for PageServerConf {
listen_pg_addr: listen_pg_addr.clone(),
listen_http_addr: listen_http_addr.clone(),
listen_https_addr: listen_https_addr.clone(),
listen_grpc_addr: listen_grpc_addr.clone(),
pg_auth_type: *pg_auth_type,
grpc_auth_type: *grpc_auth_type,
http_auth_type: *http_auth_type,
no_sync: *no_sync,
}
@@ -717,10 +707,8 @@ impl LocalEnv {
listen_pg_addr: String,
listen_http_addr: String,
listen_https_addr: Option<String>,
listen_grpc_addr: Option<String>,
pg_auth_type: AuthType,
http_auth_type: AuthType,
grpc_auth_type: AuthType,
#[serde(default)]
no_sync: bool,
}
@@ -744,10 +732,8 @@ impl LocalEnv {
listen_pg_addr,
listen_http_addr,
listen_https_addr,
listen_grpc_addr,
pg_auth_type,
http_auth_type,
grpc_auth_type,
no_sync,
} = config_toml;
let IdentityTomlSubset {
@@ -764,10 +750,8 @@ impl LocalEnv {
listen_pg_addr,
listen_http_addr,
listen_https_addr,
listen_grpc_addr,
pg_auth_type,
http_auth_type,
grpc_auth_type,
no_sync,
};
pageservers.push(conf);

View File

@@ -129,9 +129,7 @@ impl PageServerNode {
));
}
if [conf.http_auth_type, conf.pg_auth_type, conf.grpc_auth_type]
.contains(&AuthType::NeonJWT)
{
if conf.http_auth_type != AuthType::Trust || conf.pg_auth_type != AuthType::Trust {
// Keys are generated in the toplevel repo dir, pageservers' workdirs
// are one level below that, so refer to keys with ../
overrides.push("auth_validation_public_key_path='../auth_public_key.pem'".to_owned());
@@ -548,16 +546,6 @@ impl PageServerNode {
.map(serde_json::from_str)
.transpose()
.context("Falied to parse 'sampling_ratio'")?,
relsize_snapshot_cache_capacity: settings
.remove("relsize snapshot cache capacity")
.map(|x| x.parse::<usize>())
.transpose()
.context("Falied to parse 'relsize_snapshot_cache_capacity' as integer")?,
basebackup_cache_enabled: settings
.remove("basebackup_cache_enabled")
.map(|x| x.parse::<bool>())
.transpose()
.context("Failed to parse 'basebackup_cache_enabled' as bool")?,
};
if !settings.is_empty() {
bail!("Unrecognized tenant settings: {settings:?}")

View File

@@ -20,7 +20,7 @@ first_path="$(ldconfig --verbose 2>/dev/null \
| grep --invert-match ^$'\t' \
| cut --delimiter=: --fields=1 \
| head --lines=1)"
test "$first_path" == '/usr/local/lib'
test "$first_path" == '/usr/local/lib' || true # Remove the || true in a follow-up PR. Needed for backwards compat.
echo "Waiting pageserver become ready."
while ! nc -z pageserver 6400; do

View File

@@ -1,8 +0,0 @@
#!/bin/bash
# We need these settings to get the expected output results.
# We cannot use the environment variables e.g. PGTZ due to
# https://github.com/neondatabase/neon/issues/1287
export DATABASE=${1:-contrib_regression}
psql -c "ALTER DATABASE ${DATABASE} SET neon.allow_unstable_extensions='on'" \
-c "ALTER DATABASE ${DATABASE} SET DateStyle='Postgres,MDY'" \
-c "ALTER DATABASE ${DATABASE} SET TimeZone='America/Los_Angeles'" \

View File

@@ -1,16 +0,0 @@
#!/usr/bin/env bash
set -ex
cd "$(dirname "${0}")"
PG_REGRESS=$(dirname "$(pg_config --pgxs)")/../test/regress/pg_regress
dropdb --if-exists contrib_regression
createdb contrib_regression
cd h3_postgis/test
psql -d contrib_regression -c "CREATE EXTENSION postgis" -c "CREATE EXTENSION postgis_raster" -c "CREATE EXTENSION h3" -c "CREATE EXTENSION h3_postgis"
TESTS=$(echo sql/* | sed 's|sql/||g; s|\.sql||g')
${PG_REGRESS} --use-existing --dbname contrib_regression ${TESTS}
cd ../../h3/test
TESTS=$(echo sql/* | sed 's|sql/||g; s|\.sql||g')
dropdb --if-exists contrib_regression
createdb contrib_regression
psql -d contrib_regression -c "CREATE EXTENSION h3"
${PG_REGRESS} --use-existing --dbname contrib_regression ${TESTS}

View File

@@ -1,7 +0,0 @@
#!/bin/sh
set -ex
cd "$(dirname ${0})"
PG_REGRESS=$(dirname "$(pg_config --pgxs)")/../test/regress/pg_regress
cd h3/test
TESTS=$(echo sql/* | sed 's|sql/||g; s|\.sql||g')
${PG_REGRESS} --use-existing --inputdir=./ --bindir='/usr/local/pgsql/bin' --dbname=contrib_regression ${TESTS}

View File

@@ -1,6 +0,0 @@
#!/bin/sh
set -ex
cd "$(dirname "${0}")"
if [ -f Makefile ]; then
make installcheck
fi

View File

@@ -1,9 +0,0 @@
#!/bin/sh
set -ex
cd "$(dirname ${0})"
[ -f Makefile ] || exit 0
dropdb --if-exist contrib_regression
createdb contrib_regression
PG_REGRESS=$(dirname "$(pg_config --pgxs)")/../test/regress/pg_regress
TESTS=$(echo sql/* | sed 's|sql/||g; s|\.sql||g')
${PG_REGRESS} --use-existing --inputdir=./ --bindir='/usr/local/pgsql/bin' --dbname=contrib_regression ${TESTS}

View File

@@ -18,7 +18,6 @@ TESTS=${TESTS/row_level_security/}
TESTS=${TESTS/sqli_connection/}
dropdb --if-exist contrib_regression
createdb contrib_regression
. ../alter_db.sh
psql -v ON_ERROR_STOP=1 -f test/fixtures.sql -d contrib_regression
${REGRESS} --use-existing --dbname=contrib_regression --inputdir=${TESTDIR} ${TESTS}

View File

@@ -3,7 +3,6 @@ set -ex
cd "$(dirname "${0}")"
dropdb --if-exist contrib_regression
createdb contrib_regression
. ../alter_db.sh
psql -d contrib_regression -c "CREATE EXTENSION vector" -c "CREATE EXTENSION rag"
PG_REGRESS=$(dirname "$(pg_config --pgxs)")/../test/regress/pg_regress
${PG_REGRESS} --inputdir=./ --bindir='/usr/local/pgsql/bin' --use-existing --load-extension=vector --load-extension=rag --dbname=contrib_regression basic_functions text_processing api_keys chunking_functions document_processing embedding_api_functions voyageai_functions

View File

@@ -20,6 +20,5 @@ installcheck: regression-test
regression-test:
dropdb --if-exists contrib_regression
createdb contrib_regression
../alter_db.sh
psql -d contrib_regression -c "CREATE EXTENSION $(EXTNAME)"
$(PG_REGRESS) --inputdir=. --outputdir=. --use-existing --dbname=contrib_regression $(REGRESS)

View File

@@ -3,7 +3,6 @@ set -ex
cd "$(dirname ${0})"
dropdb --if-exist contrib_regression
createdb contrib_regression
. ../alter_db.sh
PG_REGRESS=$(dirname "$(pg_config --pgxs)")/../test/regress/pg_regress
REGRESS="$(make -n installcheck | awk '{print substr($0,index($0,"init-extension"));}')"
REGRESS="${REGRESS/startup_perms/}"

View File

@@ -11,6 +11,5 @@ PG_REGRESS := $(dir $(PGXS))../../src/test/regress/pg_regress
installcheck:
dropdb --if-exists contrib_regression
createdb contrib_regression
../alter_db.sh
psql -d contrib_regression -c "CREATE EXTENSION vector" -c "CREATE EXTENSION rag_bge_small_en_v15"
$(PG_REGRESS) --use-existing --dbname=contrib_regression $(REGRESS)

View File

@@ -11,6 +11,5 @@ PG_REGRESS := $(dir $(PGXS))../../src/test/regress/pg_regress
installcheck:
dropdb --if-exists contrib_regression
createdb contrib_regression
../alter_db.sh
psql -d contrib_regression -c "CREATE EXTENSION vector" -c "CREATE EXTENSION rag_jina_reranker_v1_tiny_en"
$(PG_REGRESS) --use-existing --dbname=contrib_regression $(REGRESS)

View File

@@ -3,6 +3,5 @@ set -ex
cd "$(dirname ${0})"
dropdb --if-exist contrib_regression
createdb contrib_regression
. ../alter_db.sh
PG_REGRESS=$(dirname "$(pg_config --pgxs)")/../test/regress/pg_regress
${PG_REGRESS} --inputdir=./ --bindir='/usr/local/pgsql/bin' --use-existing --dbname=contrib_regression rum rum_hash ruminv timestamp orderby orderby_hash altorder altorder_hash limits int2 int4 int8 float4 float8 money oid time timetz date interval macaddr inet cidr text varchar char bytea bit varbit numeric rum_weight expr array

View File

@@ -82,8 +82,7 @@ EXTENSIONS='[
{"extname": "pg_ivm", "extdir": "pg_ivm-src"},
{"extname": "pgjwt", "extdir": "pgjwt-src"},
{"extname": "pgtap", "extdir": "pgtap-src"},
{"extname": "pg_repack", "extdir": "pg_repack-src"},
{"extname": "h3", "extdir": "h3-pg-src"}
{"extname": "pg_repack", "extdir": "pg_repack-src"}
]'
EXTNAMES=$(echo ${EXTENSIONS} | jq -r '.[].extname' | paste -sd ' ' -)
COMPUTE_TAG=${NEW_COMPUTE_TAG} docker compose --profile test-extensions up --quiet-pull --build -d

View File

@@ -462,8 +462,6 @@ MC4CAQAwBQYDK2VwBCIEID/Drmc1AA6U/znNRWpF3zEGegOATQxfkdWxitcOMsIH
if var(REAL_S3_ENV).is_ok() {
assert!(body.contains("remote_storage_s3_deleted_objects_total"));
}
#[cfg(target_os = "linux")]
assert!(body.contains("process_threads"));
}

View File

@@ -27,7 +27,6 @@ pub use prometheus::{
pub mod launch_timestamp;
mod wrappers;
pub use prometheus;
pub use wrappers::{CountedReader, CountedWriter};
mod hll;
pub use hll::{HyperLogLog, HyperLogLogState, HyperLogLogVec};

View File

@@ -1,13 +0,0 @@
[package]
name = "neon-shmem"
version = "0.1.0"
edition.workspace = true
license.workspace = true
[dependencies]
thiserror.workspace = true
nix.workspace=true
workspace_hack = { version = "0.1", path = "../../workspace_hack" }
[target.'cfg(target_os = "macos")'.dependencies]
tempfile = "3.14.0"

View File

@@ -1,418 +0,0 @@
//! Shared memory utilities for neon communicator
use std::num::NonZeroUsize;
use std::os::fd::{AsFd, BorrowedFd, OwnedFd};
use std::ptr::NonNull;
use std::sync::atomic::{AtomicUsize, Ordering};
use nix::errno::Errno;
use nix::sys::mman::MapFlags;
use nix::sys::mman::ProtFlags;
use nix::sys::mman::mmap as nix_mmap;
use nix::sys::mman::munmap as nix_munmap;
use nix::unistd::ftruncate as nix_ftruncate;
/// ShmemHandle represents a shared memory area that can be shared by processes over fork().
/// Unlike shared memory allocated by Postgres, this area is resizable, up to 'max_size' that's
/// specified at creation.
///
/// The area is backed by an anonymous file created with memfd_create(). The full address space for
/// 'max_size' is reserved up-front with mmap(), but whenever you call [`ShmemHandle::set_size`],
/// the underlying file is resized. Do not access the area beyond the current size. Currently, that
/// will cause the file to be expanded, but we might use mprotect() etc. to enforce that in the
/// future.
pub struct ShmemHandle {
/// memfd file descriptor
fd: OwnedFd,
max_size: usize,
// Pointer to the beginning of the shared memory area. The header is stored there.
shared_ptr: NonNull<SharedStruct>,
// Pointer to the beginning of the user data
pub data_ptr: NonNull<u8>,
}
/// This is stored at the beginning in the shared memory area.
struct SharedStruct {
max_size: usize,
/// Current size of the backing file. The high-order bit is used for the RESIZE_IN_PROGRESS flag
current_size: AtomicUsize,
}
const RESIZE_IN_PROGRESS: usize = 1 << 63;
const HEADER_SIZE: usize = std::mem::size_of::<SharedStruct>();
/// Error type returned by the ShmemHandle functions.
#[derive(thiserror::Error, Debug)]
#[error("{msg}: {errno}")]
pub struct Error {
pub msg: String,
pub errno: Errno,
}
impl Error {
fn new(msg: &str, errno: Errno) -> Error {
Error {
msg: msg.to_string(),
errno,
}
}
}
impl ShmemHandle {
/// Create a new shared memory area. To communicate between processes, the processes need to be
/// fork()'d after calling this, so that the ShmemHandle is inherited by all processes.
///
/// If the ShmemHandle is dropped, the memory is unmapped from the current process. Other
/// processes can continue using it, however.
pub fn new(name: &str, initial_size: usize, max_size: usize) -> Result<ShmemHandle, Error> {
// create the backing anonymous file.
let fd = create_backing_file(name)?;
Self::new_with_fd(fd, initial_size, max_size)
}
fn new_with_fd(
fd: OwnedFd,
initial_size: usize,
max_size: usize,
) -> Result<ShmemHandle, Error> {
// We reserve the high-order bit for the RESIZE_IN_PROGRESS flag, and the actual size
// is a little larger than this because of the SharedStruct header. Make the upper limit
// somewhat smaller than that, because with anything close to that, you'll run out of
// memory anyway.
if max_size >= 1 << 48 {
panic!("max size {} too large", max_size);
}
if initial_size > max_size {
panic!("initial size {initial_size} larger than max size {max_size}");
}
// The actual initial / max size is the one given by the caller, plus the size of
// 'SharedStruct'.
let initial_size = HEADER_SIZE + initial_size;
let max_size = NonZeroUsize::new(HEADER_SIZE + max_size).unwrap();
// Reserve address space for it with mmap
//
// TODO: Use MAP_HUGETLB if possible
let start_ptr = unsafe {
nix_mmap(
None,
max_size,
ProtFlags::PROT_READ | ProtFlags::PROT_WRITE,
MapFlags::MAP_SHARED,
&fd,
0,
)
}
.map_err(|e| Error::new("mmap failed: {e}", e))?;
// Reserve space for the initial size
enlarge_file(fd.as_fd(), initial_size as u64)?;
// Initialize the header
let shared: NonNull<SharedStruct> = start_ptr.cast();
unsafe {
shared.write(SharedStruct {
max_size: max_size.into(),
current_size: AtomicUsize::new(initial_size),
})
};
// The user data begins after the header
let data_ptr = unsafe { start_ptr.cast().add(HEADER_SIZE) };
Ok(ShmemHandle {
fd,
max_size: max_size.into(),
shared_ptr: shared,
data_ptr,
})
}
// return reference to the header
fn shared(&self) -> &SharedStruct {
unsafe { self.shared_ptr.as_ref() }
}
/// Resize the shared memory area. 'new_size' must not be larger than the 'max_size' specified
/// when creating the area.
///
/// This may only be called from one process/thread concurrently. We detect that case
/// and return an Error.
pub fn set_size(&self, new_size: usize) -> Result<(), Error> {
let new_size = new_size + HEADER_SIZE;
let shared = self.shared();
if new_size > self.max_size {
panic!(
"new size ({} is greater than max size ({})",
new_size, self.max_size
);
}
assert_eq!(self.max_size, shared.max_size);
// Lock the area by setting the bit in 'current_size'
//
// Ordering::Relaxed would probably be sufficient here, as we don't access any other memory
// and the posix_fallocate/ftruncate call is surely a synchronization point anyway. But
// since this is not performance-critical, better safe than sorry .
let mut old_size = shared.current_size.load(Ordering::Acquire);
loop {
if (old_size & RESIZE_IN_PROGRESS) != 0 {
return Err(Error::new(
"concurrent resize detected",
Errno::UnknownErrno,
));
}
match shared.current_size.compare_exchange(
old_size,
new_size,
Ordering::Acquire,
Ordering::Relaxed,
) {
Ok(_) => break,
Err(x) => old_size = x,
}
}
// Ok, we got the lock.
//
// NB: If anything goes wrong, we *must* clear the bit!
let result = {
use std::cmp::Ordering::{Equal, Greater, Less};
match new_size.cmp(&old_size) {
Less => nix_ftruncate(&self.fd, new_size as i64).map_err(|e| {
Error::new("could not shrink shmem segment, ftruncate failed: {e}", e)
}),
Equal => Ok(()),
Greater => enlarge_file(self.fd.as_fd(), new_size as u64),
}
};
// Unlock
shared.current_size.store(
if result.is_ok() { new_size } else { old_size },
Ordering::Release,
);
result
}
/// Returns the current user-visible size of the shared memory segment.
///
/// NOTE: a concurrent set_size() call can change the size at any time. It is the caller's
/// responsibility not to access the area beyond the current size.
pub fn current_size(&self) -> usize {
let total_current_size =
self.shared().current_size.load(Ordering::Relaxed) & !RESIZE_IN_PROGRESS;
total_current_size - HEADER_SIZE
}
}
impl Drop for ShmemHandle {
fn drop(&mut self) {
// SAFETY: The pointer was obtained from mmap() with the given size.
// We unmap the entire region.
let _ = unsafe { nix_munmap(self.shared_ptr.cast(), self.max_size) };
// The fd is dropped automatically by OwnedFd.
}
}
/// Create a "backing file" for the shared memory area. On Linux, use memfd_create(), to create an
/// anonymous in-memory file. One macos, fall back to a regular file. That's good enough for
/// development and testing, but in production we want the file to stay in memory.
///
/// disable 'unused_variables' warnings, because in the macos path, 'name' is unused.
#[allow(unused_variables)]
fn create_backing_file(name: &str) -> Result<OwnedFd, Error> {
#[cfg(not(target_os = "macos"))]
{
nix::sys::memfd::memfd_create(name, nix::sys::memfd::MFdFlags::empty())
.map_err(|e| Error::new("memfd_create failed: {e}", e))
}
#[cfg(target_os = "macos")]
{
let file = tempfile::tempfile().map_err(|e| {
Error::new(
"could not create temporary file to back shmem area: {e}",
nix::errno::Errno::from_raw(e.raw_os_error().unwrap_or(0)),
)
})?;
Ok(OwnedFd::from(file))
}
}
fn enlarge_file(fd: BorrowedFd, size: u64) -> Result<(), Error> {
// Use posix_fallocate() to enlarge the file. It reserves the space correctly, so that
// we don't get a segfault later when trying to actually use it.
#[cfg(not(target_os = "macos"))]
{
nix::fcntl::posix_fallocate(fd, 0, size as i64).map_err(|e| {
Error::new(
"could not grow shmem segment, posix_fallocate failed: {e}",
e,
)
})
}
// As a fallback on macos, which doesn't have posix_fallocate, use plain 'fallocate'
#[cfg(target_os = "macos")]
{
nix::unistd::ftruncate(fd, size as i64)
.map_err(|e| Error::new("could not grow shmem segment, ftruncate failed: {e}", e))
}
}
#[cfg(test)]
mod tests {
use super::*;
use nix::unistd::ForkResult;
use std::ops::Range;
/// check that all bytes in given range have the expected value.
fn assert_range(ptr: *const u8, expected: u8, range: Range<usize>) {
for i in range {
let b = unsafe { *(ptr.add(i)) };
assert_eq!(expected, b, "unexpected byte at offset {}", i);
}
}
/// Write 'b' to all bytes in the given range
fn write_range(ptr: *mut u8, b: u8, range: Range<usize>) {
unsafe { std::ptr::write_bytes(ptr.add(range.start), b, range.end - range.start) };
}
// simple single-process test of growing and shrinking
#[test]
fn test_shmem_resize() -> Result<(), Error> {
let max_size = 1024 * 1024;
let init_struct = ShmemHandle::new("test_shmem_resize", 0, max_size)?;
assert_eq!(init_struct.current_size(), 0);
// Initial grow
let size1 = 10000;
init_struct.set_size(size1).unwrap();
assert_eq!(init_struct.current_size(), size1);
// Write some data
let data_ptr = init_struct.data_ptr.as_ptr();
write_range(data_ptr, 0xAA, 0..size1);
assert_range(data_ptr, 0xAA, 0..size1);
// Shrink
let size2 = 5000;
init_struct.set_size(size2).unwrap();
assert_eq!(init_struct.current_size(), size2);
// Grow again
let size3 = 20000;
init_struct.set_size(size3).unwrap();
assert_eq!(init_struct.current_size(), size3);
// Try to read it. The area that was shrunk and grown again should read as all zeros now
assert_range(data_ptr, 0xAA, 0..5000);
assert_range(data_ptr, 0, 5000..size1);
// Try to grow beyond max_size
//let size4 = max_size + 1;
//assert!(init_struct.set_size(size4).is_err());
// Dropping init_struct should unmap the memory
drop(init_struct);
Ok(())
}
/// This is used in tests to coordinate between test processes. It's like std::sync::Barrier,
/// but is stored in the shared memory area and works across processes. It's implemented by
/// polling, because e.g. standard rust mutexes are not guaranteed to work across processes.
struct SimpleBarrier {
num_procs: usize,
count: AtomicUsize,
}
impl SimpleBarrier {
unsafe fn init(ptr: *mut SimpleBarrier, num_procs: usize) {
unsafe {
*ptr = SimpleBarrier {
num_procs,
count: AtomicUsize::new(0),
}
}
}
pub fn wait(&self) {
let old = self.count.fetch_add(1, Ordering::Relaxed);
let generation = old / self.num_procs;
let mut current = old + 1;
while current < (generation + 1) * self.num_procs {
std::thread::sleep(std::time::Duration::from_millis(10));
current = self.count.load(Ordering::Relaxed);
}
}
}
#[test]
fn test_multi_process() {
// Initialize
let max_size = 1_000_000_000_000;
let init_struct = ShmemHandle::new("test_multi_process", 0, max_size).unwrap();
let ptr = init_struct.data_ptr.as_ptr();
// Store the SimpleBarrier in the first 1k of the area.
init_struct.set_size(10000).unwrap();
let barrier_ptr: *mut SimpleBarrier = unsafe {
ptr.add(ptr.align_offset(std::mem::align_of::<SimpleBarrier>()))
.cast()
};
unsafe { SimpleBarrier::init(barrier_ptr, 2) };
let barrier = unsafe { barrier_ptr.as_ref().unwrap() };
// Fork another test process. The code after this runs in both processes concurrently.
let fork_result = unsafe { nix::unistd::fork().unwrap() };
// In the parent, fill bytes between 1000..2000. In the child, between 2000..3000
if fork_result.is_parent() {
write_range(ptr, 0xAA, 1000..2000);
} else {
write_range(ptr, 0xBB, 2000..3000);
}
barrier.wait();
// Verify the contents. (in both processes)
assert_range(ptr, 0xAA, 1000..2000);
assert_range(ptr, 0xBB, 2000..3000);
// Grow, from the child this time
let size = 10_000_000;
if !fork_result.is_parent() {
init_struct.set_size(size).unwrap();
}
barrier.wait();
// make some writes at the end
if fork_result.is_parent() {
write_range(ptr, 0xAA, (size - 10)..size);
} else {
write_range(ptr, 0xBB, (size - 20)..(size - 10));
}
barrier.wait();
// Verify the contents. (This runs in both processes)
assert_range(ptr, 0, (size - 1000)..(size - 20));
assert_range(ptr, 0xBB, (size - 20)..(size - 10));
assert_range(ptr, 0xAA, (size - 10)..size);
if let ForkResult::Parent { child } = fork_result {
nix::sys::wait::waitpid(child, None).unwrap();
}
}
}

View File

@@ -8,8 +8,6 @@ pub const DEFAULT_PG_LISTEN_PORT: u16 = 64000;
pub const DEFAULT_PG_LISTEN_ADDR: &str = formatcp!("127.0.0.1:{DEFAULT_PG_LISTEN_PORT}");
pub const DEFAULT_HTTP_LISTEN_PORT: u16 = 9898;
pub const DEFAULT_HTTP_LISTEN_ADDR: &str = formatcp!("127.0.0.1:{DEFAULT_HTTP_LISTEN_PORT}");
// TODO: gRPC is disabled by default for now, but the port is used in neon_local.
pub const DEFAULT_GRPC_LISTEN_PORT: u16 = 51051; // storage-broker already uses 50051
use std::collections::HashMap;
use std::num::{NonZeroU64, NonZeroUsize};
@@ -45,21 +43,6 @@ pub struct NodeMetadata {
pub other: HashMap<String, serde_json::Value>,
}
/// PostHog integration config.
#[derive(Debug, Clone, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
pub struct PostHogConfig {
/// PostHog project ID
pub project_id: String,
/// Server-side (private) API key
pub server_api_key: String,
/// Client-side (public) API key
pub client_api_key: String,
/// Private API URL
pub private_api_url: String,
/// Public API URL
pub public_api_url: String,
}
/// `pageserver.toml`
///
/// We use serde derive with `#[serde(default)]` to generate a deserializer
@@ -121,7 +104,6 @@ pub struct ConfigToml {
pub listen_pg_addr: String,
pub listen_http_addr: String,
pub listen_https_addr: Option<String>,
pub listen_grpc_addr: Option<String>,
pub ssl_key_file: Utf8PathBuf,
pub ssl_cert_file: Utf8PathBuf,
#[serde(with = "humantime_serde")]
@@ -141,7 +123,6 @@ pub struct ConfigToml {
pub http_auth_type: AuthType,
#[serde_as(as = "serde_with::DisplayFromStr")]
pub pg_auth_type: AuthType,
pub grpc_auth_type: AuthType,
pub auth_validation_public_key_path: Option<Utf8PathBuf>,
pub remote_storage: Option<RemoteStorageConfig>,
pub tenant_config: TenantConfigToml,
@@ -201,11 +182,7 @@ pub struct ConfigToml {
pub tracing: Option<Tracing>,
pub enable_tls_page_service_api: bool,
pub dev_mode: bool,
#[serde(skip_serializing_if = "Option::is_none")]
pub posthog_config: Option<PostHogConfig>,
pub timeline_import_config: TimelineImportConfig,
#[serde(skip_serializing_if = "Option::is_none")]
pub basebackup_cache_config: Option<BasebackupCacheConfig>,
}
#[derive(Debug, Clone, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
@@ -258,7 +235,7 @@ pub enum PageServiceProtocolPipelinedBatchingStrategy {
ScatteredLsn,
}
#[derive(Debug, Clone, Copy, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
#[derive(Debug, Clone, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
#[serde(tag = "mode", rename_all = "kebab-case")]
pub enum GetVectoredConcurrentIo {
/// The read path is fully sequential: layers are visited
@@ -328,27 +305,6 @@ impl From<OtelExporterProtocol> for tracing_utils::Protocol {
pub struct TimelineImportConfig {
pub import_job_concurrency: NonZeroUsize,
pub import_job_soft_size_limit: NonZeroUsize,
pub import_job_checkpoint_threshold: NonZeroUsize,
}
#[derive(Debug, Clone, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
#[serde(default)]
pub struct BasebackupCacheConfig {
#[serde(with = "humantime_serde")]
pub cleanup_period: Duration,
// FIXME: Support max_size_bytes.
// pub max_size_bytes: usize,
pub max_size_entries: i64,
}
impl Default for BasebackupCacheConfig {
fn default() -> Self {
Self {
cleanup_period: Duration::from_secs(60),
// max_size_bytes: 1024 * 1024 * 1024, // 1 GiB
max_size_entries: 1000,
}
}
}
pub mod statvfs {
@@ -534,14 +490,6 @@ pub struct TenantConfigToml {
/// Tenant level performance sampling ratio override. Controls the ratio of get page requests
/// that will get perf sampling for the tenant.
pub sampling_ratio: Option<Ratio>,
/// Capacity of relsize snapshot cache (used by replicas).
pub relsize_snapshot_cache_capacity: usize,
/// Enable preparing basebackup on XLOG_CHECKPOINT_SHUTDOWN and using it in basebackup requests.
// FIXME: Remove skip_serializing_if when the feature is stable.
#[serde(skip_serializing_if = "std::ops::Not::not")]
pub basebackup_cache_enabled: bool,
}
pub mod defaults {
@@ -609,7 +557,6 @@ impl Default for ConfigToml {
listen_pg_addr: (DEFAULT_PG_LISTEN_ADDR.to_string()),
listen_http_addr: (DEFAULT_HTTP_LISTEN_ADDR.to_string()),
listen_https_addr: (None),
listen_grpc_addr: None, // TODO: default to 127.0.0.1:51051
ssl_key_file: Utf8PathBuf::from(DEFAULT_SSL_KEY_FILE),
ssl_cert_file: Utf8PathBuf::from(DEFAULT_SSL_CERT_FILE),
ssl_cert_reload_period: Duration::from_secs(60),
@@ -626,7 +573,6 @@ impl Default for ConfigToml {
pg_distrib_dir: None, // Utf8PathBuf::from("./pg_install"), // TODO: formely, this was std::env::current_dir()
http_auth_type: (AuthType::Trust),
pg_auth_type: (AuthType::Trust),
grpc_auth_type: (AuthType::Trust),
auth_validation_public_key_path: (None),
remote_storage: None,
broker_endpoint: (storage_broker::DEFAULT_ENDPOINT
@@ -693,15 +639,23 @@ impl Default for ConfigToml {
tenant_config: TenantConfigToml::default(),
no_sync: None,
wal_receiver_protocol: DEFAULT_WAL_RECEIVER_PROTOCOL,
page_service_pipelining: PageServicePipeliningConfig::Pipelined(
PageServicePipeliningConfigPipelined {
page_service_pipelining: if !cfg!(test) {
PageServicePipeliningConfig::Serial
} else {
// Do not turn this into the default until scattered reads have been
// validated and rolled-out fully.
PageServicePipeliningConfig::Pipelined(PageServicePipeliningConfigPipelined {
max_batch_size: NonZeroUsize::new(32).unwrap(),
execution: PageServiceProtocolPipelinedExecutionStrategy::ConcurrentFutures,
batching: PageServiceProtocolPipelinedBatchingStrategy::ScatteredLsn,
},
),
get_vectored_concurrent_io: GetVectoredConcurrentIo::SidecarTask,
enable_read_path_debugging: if cfg!(feature = "testing") {
})
},
get_vectored_concurrent_io: if !cfg!(test) {
GetVectoredConcurrentIo::Sequential
} else {
GetVectoredConcurrentIo::SidecarTask
},
enable_read_path_debugging: if cfg!(test) || cfg!(feature = "testing") {
Some(true)
} else {
None
@@ -715,10 +669,7 @@ impl Default for ConfigToml {
timeline_import_config: TimelineImportConfig {
import_job_concurrency: NonZeroUsize::new(128).unwrap(),
import_job_soft_size_limit: NonZeroUsize::new(1024 * 1024 * 1024).unwrap(),
import_job_checkpoint_threshold: NonZeroUsize::new(128).unwrap(),
},
basebackup_cache_config: None,
posthog_config: None,
}
}
}
@@ -785,7 +736,6 @@ pub mod tenant_conf_defaults {
pub const DEFAULT_GC_COMPACTION_VERIFICATION: bool = true;
pub const DEFAULT_GC_COMPACTION_INITIAL_THRESHOLD_KB: u64 = 5 * 1024 * 1024; // 5GB
pub const DEFAULT_GC_COMPACTION_RATIO_PERCENT: u64 = 100;
pub const DEFAULT_RELSIZE_SNAPSHOT_CACHE_CAPACITY: usize = 1000;
}
impl Default for TenantConfigToml {
@@ -843,8 +793,6 @@ impl Default for TenantConfigToml {
gc_compaction_initial_threshold_kb: DEFAULT_GC_COMPACTION_INITIAL_THRESHOLD_KB,
gc_compaction_ratio_percent: DEFAULT_GC_COMPACTION_RATIO_PERCENT,
sampling_ratio: None,
relsize_snapshot_cache_capacity: DEFAULT_RELSIZE_SNAPSHOT_CACHE_CAPACITY,
basebackup_cache_enabled: false,
}
}
}

View File

@@ -910,11 +910,6 @@ impl Key {
self.field1 == 0x00 && self.field4 != 0 && self.field6 != 0xffffffff
}
#[inline(always)]
pub fn is_rel_block_of_rel(&self, rel: Oid) -> bool {
self.is_rel_block_key() && self.field4 == rel
}
#[inline(always)]
pub fn is_rel_dir_key(&self) -> bool {
self.field1 == 0x00

View File

@@ -336,33 +336,14 @@ impl TimelineCreateRequest {
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Eq)]
pub enum ShardImportStatus {
InProgress(Option<ShardImportProgress>),
InProgress,
Done,
Error(String),
}
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Eq)]
pub enum ShardImportProgress {
V1(ShardImportProgressV1),
}
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Eq)]
pub struct ShardImportProgressV1 {
/// Total number of jobs in the import plan
pub jobs: usize,
/// Number of jobs completed
pub completed: usize,
/// Hash of the plan
pub import_plan_hash: u64,
/// Soft limit for the job size
/// This needs to remain constant throughout the import
pub job_soft_size_limit: usize,
}
impl ShardImportStatus {
pub fn is_terminal(&self) -> bool {
match self {
ShardImportStatus::InProgress(_) => false,
ShardImportStatus::InProgress => false,
ShardImportStatus::Done | ShardImportStatus::Error(_) => true,
}
}
@@ -405,8 +386,6 @@ pub enum TimelineCreateRequestMode {
// using a flattened enum, so, it was an accepted field, and
// we continue to accept it by having it here.
pg_version: Option<u32>,
#[serde(default, skip_serializing_if = "std::ops::Not::not")]
read_only: bool,
},
ImportPgdata {
import_pgdata: TimelineCreateRequestModeImportPgdata,
@@ -635,10 +614,6 @@ pub struct TenantConfigPatch {
pub gc_compaction_ratio_percent: FieldPatch<u64>,
#[serde(skip_serializing_if = "FieldPatch::is_noop")]
pub sampling_ratio: FieldPatch<Option<Ratio>>,
#[serde(skip_serializing_if = "FieldPatch::is_noop")]
pub relsize_snapshot_cache_capacity: FieldPatch<usize>,
#[serde(skip_serializing_if = "FieldPatch::is_noop")]
pub basebackup_cache_enabled: FieldPatch<bool>,
}
/// Like [`crate::config::TenantConfigToml`], but preserves the information
@@ -768,12 +743,6 @@ pub struct TenantConfig {
#[serde(skip_serializing_if = "Option::is_none")]
pub sampling_ratio: Option<Option<Ratio>>,
#[serde(skip_serializing_if = "Option::is_none")]
pub relsize_snapshot_cache_capacity: Option<usize>,
#[serde(skip_serializing_if = "Option::is_none")]
pub basebackup_cache_enabled: Option<bool>,
}
impl TenantConfig {
@@ -819,8 +788,6 @@ impl TenantConfig {
mut gc_compaction_initial_threshold_kb,
mut gc_compaction_ratio_percent,
mut sampling_ratio,
mut relsize_snapshot_cache_capacity,
mut basebackup_cache_enabled,
} = self;
patch.checkpoint_distance.apply(&mut checkpoint_distance);
@@ -922,12 +889,6 @@ impl TenantConfig {
.gc_compaction_ratio_percent
.apply(&mut gc_compaction_ratio_percent);
patch.sampling_ratio.apply(&mut sampling_ratio);
patch
.relsize_snapshot_cache_capacity
.apply(&mut relsize_snapshot_cache_capacity);
patch
.basebackup_cache_enabled
.apply(&mut basebackup_cache_enabled);
Ok(Self {
checkpoint_distance,
@@ -967,8 +928,6 @@ impl TenantConfig {
gc_compaction_initial_threshold_kb,
gc_compaction_ratio_percent,
sampling_ratio,
relsize_snapshot_cache_capacity,
basebackup_cache_enabled,
})
}
@@ -1077,12 +1036,6 @@ impl TenantConfig {
.gc_compaction_ratio_percent
.unwrap_or(global_conf.gc_compaction_ratio_percent),
sampling_ratio: self.sampling_ratio.unwrap_or(global_conf.sampling_ratio),
relsize_snapshot_cache_capacity: self
.relsize_snapshot_cache_capacity
.unwrap_or(global_conf.relsize_snapshot_cache_capacity),
basebackup_cache_enabled: self
.basebackup_cache_enabled
.unwrap_or(global_conf.basebackup_cache_enabled),
}
}
}
@@ -1850,6 +1803,7 @@ pub struct TopTenantShardsResponse {
}
pub mod virtual_file {
use std::sync::LazyLock;
#[derive(
Copy,
@@ -1897,7 +1851,15 @@ pub mod virtual_file {
impl IoMode {
pub fn preferred() -> Self {
IoMode::DirectRw
// The default behavior when running Rust unit tests without any further
// flags is to use the newest behavior (DirectRw).
// The CI uses the environment variable to unit tests for all different modes.
// NB: the Python regression & perf tests have their own defaults management
// that writes pageserver.toml; they do not use this variable.
static ENV_OVERRIDE: LazyLock<Option<IoMode>> = LazyLock::new(|| {
utils::env::var_serde_json_string("NEON_PAGESERVER_UNIT_TEST_VIRTUAL_FILE_IO_MODE")
});
ENV_OVERRIDE.unwrap_or(IoMode::DirectRw)
}
}

View File

@@ -4,7 +4,6 @@
//! See docs/rfcs/025-generation-numbers.md
use serde::{Deserialize, Serialize};
use utils::generation::Generation;
use utils::id::{NodeId, TimelineId};
use crate::controller_api::NodeRegisterRequest;
@@ -64,17 +63,9 @@ pub struct ValidateResponseTenant {
pub valid: bool,
}
#[derive(Serialize, Deserialize)]
pub struct TimelineImportStatusRequest {
pub tenant_shard_id: TenantShardId,
pub timeline_id: TimelineId,
pub generation: Generation,
}
#[derive(Serialize, Deserialize)]
pub struct PutTimelineImportStatusRequest {
pub tenant_shard_id: TenantShardId,
pub timeline_id: TimelineId,
pub status: ShardImportStatus,
pub generation: Generation,
}

View File

@@ -6,14 +6,9 @@ license.workspace = true
[dependencies]
anyhow.workspace = true
arc-swap.workspace = true
reqwest.workspace = true
serde_json.workspace = true
serde.workspace = true
serde_json.workspace = true
sha2.workspace = true
thiserror.workspace = true
tokio = { workspace = true, features = ["process", "sync", "fs", "rt", "io-util", "time"] }
tokio-util.workspace = true
tracing-utils.workspace = true
tracing.workspace = true
workspace_hack.workspace = true
thiserror.workspace = true

View File

@@ -1,64 +0,0 @@
//! A background loop that fetches feature flags from PostHog and updates the feature store.
use std::{sync::Arc, time::Duration};
use arc_swap::ArcSwap;
use tokio_util::sync::CancellationToken;
use tracing::{Instrument, info_span};
use crate::{FeatureStore, PostHogClient, PostHogClientConfig};
/// A background loop that fetches feature flags from PostHog and updates the feature store.
pub struct FeatureResolverBackgroundLoop {
posthog_client: PostHogClient,
feature_store: ArcSwap<FeatureStore>,
cancel: CancellationToken,
}
impl FeatureResolverBackgroundLoop {
pub fn new(config: PostHogClientConfig, shutdown_pageserver: CancellationToken) -> Self {
Self {
posthog_client: PostHogClient::new(config),
feature_store: ArcSwap::new(Arc::new(FeatureStore::new())),
cancel: shutdown_pageserver,
}
}
pub fn spawn(self: Arc<Self>, handle: &tokio::runtime::Handle, refresh_period: Duration) {
let this = self.clone();
let cancel = self.cancel.clone();
handle.spawn(
async move {
tracing::info!("Starting PostHog feature resolver");
let mut ticker = tokio::time::interval(refresh_period);
ticker.set_missed_tick_behavior(tokio::time::MissedTickBehavior::Skip);
loop {
tokio::select! {
_ = ticker.tick() => {}
_ = cancel.cancelled() => break
}
let resp = match this
.posthog_client
.get_feature_flags_local_evaluation()
.await
{
Ok(resp) => resp,
Err(e) => {
tracing::warn!("Cannot get feature flags: {}", e);
continue;
}
};
let feature_store = FeatureStore::new_with_flags(resp.flags);
this.feature_store.store(Arc::new(feature_store));
tracing::info!("Feature flag updated");
}
tracing::info!("PostHog feature resolver stopped");
}
.instrument(info_span!("posthog_feature_resolver")),
);
}
pub fn feature_store(&self) -> Arc<FeatureStore> {
self.feature_store.load_full()
}
}

View File

@@ -1,9 +1,5 @@
//! A lite version of the PostHog client that only supports local evaluation of feature flags.
mod background_loop;
pub use background_loop::FeatureResolverBackgroundLoop;
use std::collections::HashMap;
use serde::{Deserialize, Serialize};
@@ -24,7 +20,8 @@ pub enum PostHogEvaluationError {
#[derive(Deserialize)]
pub struct LocalEvaluationResponse {
pub flags: Vec<LocalEvaluationFlag>,
#[allow(dead_code)]
flags: Vec<LocalEvaluationFlag>,
}
#[derive(Deserialize)]
@@ -37,7 +34,7 @@ pub struct LocalEvaluationFlag {
#[derive(Deserialize)]
pub struct LocalEvaluationFlagFilters {
groups: Vec<LocalEvaluationFlagFilterGroup>,
multivariate: Option<LocalEvaluationFlagMultivariate>,
multivariate: LocalEvaluationFlagMultivariate,
}
#[derive(Deserialize)]
@@ -97,12 +94,6 @@ impl FeatureStore {
}
}
pub fn new_with_flags(flags: Vec<LocalEvaluationFlag>) -> Self {
let mut store = Self::new();
store.set_flags(flags);
store
}
pub fn set_flags(&mut self, flags: Vec<LocalEvaluationFlag>) {
self.flags.clear();
for flag in flags {
@@ -254,7 +245,7 @@ impl FeatureStore {
}
}
/// Evaluate a multivariate feature flag. Returns an error if the flag is not available or if there are errors
/// Evaluate a multivariate feature flag. Returns `None` if the flag is not available or if there are errors
/// during the evaluation.
///
/// The parsing logic is as follows:
@@ -272,15 +263,10 @@ impl FeatureStore {
/// Example: we have a multivariate flag with 3 groups of the configured global rollout percentage: A (10%), B (20%), C (70%).
/// There is a single group with a condition that has a rollout percentage of 10% and it does not have a variant override.
/// Then, we will have 1% of the users evaluated to A, 2% to B, and 7% to C.
///
/// Error handling: the caller should inspect the error and decide the behavior when a feature flag
/// cannot be evaluated (i.e., default to false if it cannot be resolved). The error should *not* be
/// propagated beyond where the feature flag gets resolved.
pub fn evaluate_multivariate(
&self,
flag_key: &str,
user_id: &str,
properties: &HashMap<String, PostHogFlagFilterPropertyValue>,
) -> Result<String, PostHogEvaluationError> {
let hash_on_global_rollout_percentage =
Self::consistent_hash(user_id, flag_key, "multivariate");
@@ -290,39 +276,10 @@ impl FeatureStore {
flag_key,
hash_on_global_rollout_percentage,
hash_on_group_rollout_percentage,
properties,
&HashMap::new(),
)
}
/// Evaluate a boolean feature flag. Returns an error if the flag is not available or if there are errors
/// during the evaluation.
///
/// The parsing logic is as follows:
///
/// * Generate a consistent hash for the tenant-feature.
/// * Match each filter group.
/// - If a group is matched, it will first determine whether the user is in the range of the rollout
/// percentage.
/// - If the hash falls within the group's rollout percentage, return true.
/// * Otherwise, continue with the next group until all groups are evaluated and no group is within the
/// rollout percentage.
/// * If there are no matching groups, return an error.
///
/// Returns `Ok(())` if the feature flag evaluates to true. In the future, it will return a payload.
///
/// Error handling: the caller should inspect the error and decide the behavior when a feature flag
/// cannot be evaluated (i.e., default to false if it cannot be resolved). The error should *not* be
/// propagated beyond where the feature flag gets resolved.
pub fn evaluate_boolean(
&self,
flag_key: &str,
user_id: &str,
properties: &HashMap<String, PostHogFlagFilterPropertyValue>,
) -> Result<(), PostHogEvaluationError> {
let hash_on_global_rollout_percentage = Self::consistent_hash(user_id, flag_key, "boolean");
self.evaluate_boolean_inner(flag_key, hash_on_global_rollout_percentage, properties)
}
/// Evaluate a multivariate feature flag. Note that we directly take the mapped user ID
/// (a consistent hash ranging from 0 to 1) so that it is easier to use it in the tests
/// and avoid duplicate computations.
@@ -349,11 +306,6 @@ impl FeatureStore {
flag_key
)));
}
let Some(ref multivariate) = flag_config.filters.multivariate else {
return Err(PostHogEvaluationError::Internal(format!(
"No multivariate available, should use evaluate_boolean?: {flag_key}"
)));
};
// TODO: sort the groups so that variant overrides always get evaluated first and it follows the PostHog
// Python SDK behavior; for now we do not configure conditions without variant overrides in Neon so it
// does not matter.
@@ -362,7 +314,7 @@ impl FeatureStore {
GroupEvaluationResult::MatchedAndOverride(variant) => return Ok(variant),
GroupEvaluationResult::MatchedAndEvaluate => {
let mut percentage = 0;
for variant in &multivariate.variants {
for variant in &flag_config.filters.multivariate.variants {
percentage += variant.rollout_percentage;
if self
.evaluate_percentage(hash_on_global_rollout_percentage, percentage)
@@ -390,89 +342,6 @@ impl FeatureStore {
)))
}
}
/// Evaluate a multivariate feature flag. Note that we directly take the mapped user ID
/// (a consistent hash ranging from 0 to 1) so that it is easier to use it in the tests
/// and avoid duplicate computations.
///
/// Use a different consistent hash for evaluating the group rollout percentage.
/// The behavior: if the condition is set to rolling out to 10% of the users, and
/// we set the variant A to 20% in the global config, then 2% of the total users will
/// be evaluated to variant A.
///
/// Note that the hash to determine group rollout percentage is shared across all groups. So if we have two
/// exactly-the-same conditions with 10% and 20% rollout percentage respectively, a total of 20% of the users
/// will be evaluated (versus 30% if group evaluation is done independently).
pub(crate) fn evaluate_boolean_inner(
&self,
flag_key: &str,
hash_on_global_rollout_percentage: f64,
properties: &HashMap<String, PostHogFlagFilterPropertyValue>,
) -> Result<(), PostHogEvaluationError> {
if let Some(flag_config) = self.flags.get(flag_key) {
if !flag_config.active {
return Err(PostHogEvaluationError::NotAvailable(format!(
"The feature flag is not active: {}",
flag_key
)));
}
if flag_config.filters.multivariate.is_some() {
return Err(PostHogEvaluationError::Internal(format!(
"This looks like a multivariate flag, should use evaluate_multivariate?: {flag_key}"
)));
};
// TODO: sort the groups so that variant overrides always get evaluated first and it follows the PostHog
// Python SDK behavior; for now we do not configure conditions without variant overrides in Neon so it
// does not matter.
for group in &flag_config.filters.groups {
match self.evaluate_group(group, hash_on_global_rollout_percentage, properties)? {
GroupEvaluationResult::MatchedAndOverride(_) => {
return Err(PostHogEvaluationError::Internal(format!(
"Boolean flag cannot have overrides: {}",
flag_key
)));
}
GroupEvaluationResult::MatchedAndEvaluate => {
return Ok(());
}
GroupEvaluationResult::Unmatched => continue,
}
}
// If no group is matched, the feature is not available, and up to the caller to decide what to do.
Err(PostHogEvaluationError::NoConditionGroupMatched)
} else {
// The feature flag is not available yet
Err(PostHogEvaluationError::NotAvailable(format!(
"Not found in the local evaluation spec: {}",
flag_key
)))
}
}
/// Infer whether a feature flag is a boolean flag by checking if it has a multivariate filter.
pub fn is_feature_flag_boolean(&self, flag_key: &str) -> Result<bool, PostHogEvaluationError> {
if let Some(flag_config) = self.flags.get(flag_key) {
Ok(flag_config.filters.multivariate.is_none())
} else {
Err(PostHogEvaluationError::NotAvailable(format!(
"Not found in the local evaluation spec: {}",
flag_key
)))
}
}
}
pub struct PostHogClientConfig {
/// The server API key.
pub server_api_key: String,
/// The client API key.
pub client_api_key: String,
/// The project ID.
pub project_id: String,
/// The private API URL.
pub private_api_url: String,
/// The public API URL.
pub public_api_url: String,
}
/// A lite PostHog client.
@@ -491,16 +360,37 @@ pub struct PostHogClientConfig {
/// want to report the feature flag usage back to PostHog. The current plan is to use PostHog only as an UI to
/// configure feature flags so it is very likely that the client API will not be used.
pub struct PostHogClient {
/// The config.
config: PostHogClientConfig,
/// The server API key.
server_api_key: String,
/// The client API key.
client_api_key: String,
/// The project ID.
project_id: String,
/// The private API URL.
private_api_url: String,
/// The public API URL.
public_api_url: String,
/// The HTTP client.
client: reqwest::Client,
}
impl PostHogClient {
pub fn new(config: PostHogClientConfig) -> Self {
pub fn new(
server_api_key: String,
client_api_key: String,
project_id: String,
private_api_url: String,
public_api_url: String,
) -> Self {
let client = reqwest::Client::new();
Self { config, client }
Self {
server_api_key,
client_api_key,
project_id,
private_api_url,
public_api_url,
client,
}
}
pub fn new_with_us_region(
@@ -508,13 +398,13 @@ impl PostHogClient {
client_api_key: String,
project_id: String,
) -> Self {
Self::new(PostHogClientConfig {
Self::new(
server_api_key,
client_api_key,
project_id,
private_api_url: "https://us.posthog.com".to_string(),
public_api_url: "https://us.i.posthog.com".to_string(),
})
"https://us.posthog.com".to_string(),
"https://us.i.posthog.com".to_string(),
)
}
/// Fetch the feature flag specs from the server.
@@ -532,23 +422,15 @@ impl PostHogClient {
// with bearer token of self.server_api_key
let url = format!(
"{}/api/projects/{}/feature_flags/local_evaluation",
self.config.private_api_url, self.config.project_id
self.private_api_url, self.project_id
);
let response = self
.client
.get(url)
.bearer_auth(&self.config.server_api_key)
.bearer_auth(&self.server_api_key)
.send()
.await?;
let status = response.status();
let body = response.text().await?;
if !status.is_success() {
return Err(anyhow::anyhow!(
"Failed to get feature flags: {}, {}",
status,
body
));
}
Ok(serde_json::from_str(&body)?)
}
@@ -564,11 +446,11 @@ impl PostHogClient {
) -> anyhow::Result<()> {
// PUBLIC_URL/capture/
// with bearer token of self.client_api_key
let url = format!("{}/capture/", self.config.public_api_url);
let url = format!("{}/capture/", self.public_api_url);
self.client
.post(url)
.body(serde_json::to_string(&json!({
"api_key": self.config.client_api_key,
"api_key": self.client_api_key,
"distinct_id": distinct_id,
"event": event,
"properties": properties,
@@ -585,162 +467,95 @@ mod tests {
fn data() -> &'static str {
r#"{
"flags": [
{
"id": 141807,
"team_id": 152860,
"name": "",
"key": "image-compaction-boundary",
"filters": {
"groups": [
{
"variant": null,
"properties": [
{
"key": "plan_type",
"type": "person",
"value": [
"free"
],
"operator": "exact"
}
"flags": [
{
"id": 132794,
"team_id": 152860,
"name": "",
"key": "gc-compaction",
"filters": {
"groups": [
{
"variant": "enabled-stage-2",
"properties": [
{
"key": "plan_type",
"type": "person",
"value": [
"free"
],
"operator": "exact"
},
{
"key": "pageserver_remote_size",
"type": "person",
"value": "10000000",
"operator": "lt"
}
],
"rollout_percentage": 50
},
{
"properties": [
{
"key": "plan_type",
"type": "person",
"value": [
"free"
],
"operator": "exact"
},
{
"key": "pageserver_remote_size",
"type": "person",
"value": "10000000",
"operator": "lt"
}
],
"rollout_percentage": 80
}
],
"payloads": {},
"multivariate": {
"variants": [
{
"key": "disabled",
"name": "",
"rollout_percentage": 90
},
{
"key": "enabled-stage-1",
"name": "",
"rollout_percentage": 10
},
{
"key": "enabled-stage-2",
"name": "",
"rollout_percentage": 0
},
{
"key": "enabled-stage-3",
"name": "",
"rollout_percentage": 0
},
{
"key": "enabled",
"name": "",
"rollout_percentage": 0
}
]
}
},
"deleted": false,
"active": true,
"ensure_experience_continuity": false,
"has_encrypted_payloads": false,
"version": 6
}
],
"rollout_percentage": 40
},
{
"variant": null,
"properties": [],
"rollout_percentage": 10
}
],
"payloads": {},
"multivariate": null
},
"deleted": false,
"active": true,
"ensure_experience_continuity": false,
"has_encrypted_payloads": false,
"version": 1
},
{
"id": 135586,
"team_id": 152860,
"name": "",
"key": "boolean-flag",
"filters": {
"groups": [
{
"variant": null,
"properties": [
{
"key": "plan_type",
"type": "person",
"value": [
"free"
],
"operator": "exact"
}
],
"rollout_percentage": 47
}
],
"payloads": {},
"multivariate": null
},
"deleted": false,
"active": true,
"ensure_experience_continuity": false,
"has_encrypted_payloads": false,
"version": 1
},
{
"id": 132794,
"team_id": 152860,
"name": "",
"key": "gc-compaction",
"filters": {
"groups": [
{
"variant": "enabled-stage-2",
"properties": [
{
"key": "plan_type",
"type": "person",
"value": [
"free"
],
"operator": "exact"
},
{
"key": "pageserver_remote_size",
"type": "person",
"value": "10000000",
"operator": "lt"
}
],
"rollout_percentage": 50
},
{
"properties": [
{
"key": "plan_type",
"type": "person",
"value": [
"free"
],
"operator": "exact"
},
{
"key": "pageserver_remote_size",
"type": "person",
"value": "10000000",
"operator": "lt"
}
],
"rollout_percentage": 80
}
],
"payloads": {},
"multivariate": {
"variants": [
{
"key": "disabled",
"name": "",
"rollout_percentage": 90
},
{
"key": "enabled-stage-1",
"name": "",
"rollout_percentage": 10
},
{
"key": "enabled-stage-2",
"name": "",
"rollout_percentage": 0
},
{
"key": "enabled-stage-3",
"name": "",
"rollout_percentage": 0
},
{
"key": "enabled",
"name": "",
"rollout_percentage": 0
}
]
}
},
"deleted": false,
"active": true,
"ensure_experience_continuity": false,
"has_encrypted_payloads": false,
"version": 7
}
],
"group_type_mapping": {},
"cohorts": {}
}"#
"group_type_mapping": {},
"cohorts": {}
}"#
}
#[test]
@@ -816,125 +631,4 @@ mod tests {
Err(PostHogEvaluationError::NoConditionGroupMatched)
),);
}
#[test]
fn evaluate_boolean_1() {
// The `boolean-flag` feature flag only has one group that matches on the free user.
let mut store = FeatureStore::new();
let response: LocalEvaluationResponse = serde_json::from_str(data()).unwrap();
store.set_flags(response.flags);
// This lacks the required properties and cannot be evaluated.
let variant = store.evaluate_boolean_inner("boolean-flag", 1.00, &HashMap::new());
assert!(matches!(
variant,
Err(PostHogEvaluationError::NotAvailable(_))
),);
let properties_unmatched = HashMap::from([
(
"plan_type".to_string(),
PostHogFlagFilterPropertyValue::String("paid".to_string()),
),
(
"pageserver_remote_size".to_string(),
PostHogFlagFilterPropertyValue::Number(1000.0),
),
]);
// This does not match any group so there will be an error.
let variant = store.evaluate_boolean_inner("boolean-flag", 1.00, &properties_unmatched);
assert!(matches!(
variant,
Err(PostHogEvaluationError::NoConditionGroupMatched)
),);
let properties = HashMap::from([
(
"plan_type".to_string(),
PostHogFlagFilterPropertyValue::String("free".to_string()),
),
(
"pageserver_remote_size".to_string(),
PostHogFlagFilterPropertyValue::Number(1000.0),
),
]);
// It matches the first group as 0.10 <= 0.50 and the properties are matched. Then it gets evaluated to the variant override.
let variant = store.evaluate_boolean_inner("boolean-flag", 0.10, &properties);
assert!(variant.is_ok());
// It matches the group conditions but not the group rollout percentage.
let variant = store.evaluate_boolean_inner("boolean-flag", 1.00, &properties);
assert!(matches!(
variant,
Err(PostHogEvaluationError::NoConditionGroupMatched)
),);
}
#[test]
fn evaluate_boolean_2() {
// The `image-compaction-boundary` feature flag has one group that matches on the free user and a group that matches on all users.
let mut store = FeatureStore::new();
let response: LocalEvaluationResponse = serde_json::from_str(data()).unwrap();
store.set_flags(response.flags);
// This lacks the required properties and cannot be evaluated.
let variant =
store.evaluate_boolean_inner("image-compaction-boundary", 1.00, &HashMap::new());
assert!(matches!(
variant,
Err(PostHogEvaluationError::NotAvailable(_))
),);
let properties_unmatched = HashMap::from([
(
"plan_type".to_string(),
PostHogFlagFilterPropertyValue::String("paid".to_string()),
),
(
"pageserver_remote_size".to_string(),
PostHogFlagFilterPropertyValue::Number(1000.0),
),
]);
// This does not match the filtered group but the all user group.
let variant =
store.evaluate_boolean_inner("image-compaction-boundary", 1.00, &properties_unmatched);
assert!(matches!(
variant,
Err(PostHogEvaluationError::NoConditionGroupMatched)
),);
let variant =
store.evaluate_boolean_inner("image-compaction-boundary", 0.05, &properties_unmatched);
assert!(variant.is_ok());
let properties = HashMap::from([
(
"plan_type".to_string(),
PostHogFlagFilterPropertyValue::String("free".to_string()),
),
(
"pageserver_remote_size".to_string(),
PostHogFlagFilterPropertyValue::Number(1000.0),
),
]);
// It matches the first group as 0.30 <= 0.40 and the properties are matched. Then it gets evaluated to the variant override.
let variant = store.evaluate_boolean_inner("image-compaction-boundary", 0.30, &properties);
assert!(variant.is_ok());
// It matches the group conditions but not the group rollout percentage.
let variant = store.evaluate_boolean_inner("image-compaction-boundary", 1.00, &properties);
assert!(matches!(
variant,
Err(PostHogEvaluationError::NoConditionGroupMatched)
),);
// It matches the second "all" group conditions.
let variant = store.evaluate_boolean_inner("image-compaction-boundary", 0.09, &properties);
assert!(variant.is_ok());
}
}

View File

@@ -25,7 +25,6 @@ where
Ok(())
}
#[derive(Debug)]
pub enum BindError {
Conversion(Box<dyn Error + marker::Sync + Send>),
Serialization(io::Error),
@@ -289,12 +288,6 @@ pub fn sync(buf: &mut BytesMut) {
write_body(buf, |_| Ok::<(), io::Error>(())).unwrap();
}
#[inline]
pub fn flush(buf: &mut BytesMut) {
buf.put_u8(b'H');
write_body(buf, |_| Ok::<(), io::Error>(())).unwrap();
}
#[inline]
pub fn terminate(buf: &mut BytesMut) {
buf.put_u8(b'X');

View File

@@ -9,6 +9,7 @@ use std::error::Error;
use std::fmt;
use std::sync::Arc;
use bytes::BytesMut;
use fallible_iterator::FallibleIterator;
#[doc(inline)]
pub use postgres_protocol2::Oid;
@@ -26,6 +27,41 @@ macro_rules! accepts {
)
}
/// Generates an implementation of `ToSql::to_sql_checked`.
///
/// All `ToSql` implementations should use this macro.
macro_rules! to_sql_checked {
() => {
fn to_sql_checked(
&self,
ty: &$crate::Type,
out: &mut $crate::private::BytesMut,
) -> ::std::result::Result<
$crate::IsNull,
Box<dyn ::std::error::Error + ::std::marker::Sync + ::std::marker::Send>,
> {
$crate::__to_sql_checked(self, ty, out)
}
};
}
// WARNING: this function is not considered part of this crate's public API.
// It is subject to change at any time.
#[doc(hidden)]
pub fn __to_sql_checked<T>(
v: &T,
ty: &Type,
out: &mut BytesMut,
) -> Result<IsNull, Box<dyn Error + Sync + Send>>
where
T: ToSql,
{
if !T::accepts(ty) {
return Err(Box::new(WrongType::new::<T>(ty.clone())));
}
v.to_sql(ty, out)
}
// mod pg_lsn;
#[doc(hidden)]
pub mod private;
@@ -106,7 +142,7 @@ pub enum Kind {
/// An array type along with the type of its elements.
Array(Type),
/// A range type along with the type of its elements.
Range(Oid),
Range(Type),
/// A multirange type along with the type of its elements.
Multirange(Type),
/// A domain type along with its underlying type.
@@ -341,6 +377,43 @@ pub enum IsNull {
No,
}
/// A trait for types that can be converted into Postgres values.
pub trait ToSql: fmt::Debug {
/// Converts the value of `self` into the binary format of the specified
/// Postgres `Type`, appending it to `out`.
///
/// The caller of this method is responsible for ensuring that this type
/// is compatible with the Postgres `Type`.
///
/// The return value indicates if this value should be represented as
/// `NULL`. If this is the case, implementations **must not** write
/// anything to `out`.
fn to_sql(&self, ty: &Type, out: &mut BytesMut) -> Result<IsNull, Box<dyn Error + Sync + Send>>
where
Self: Sized;
/// Determines if a value of this type can be converted to the specified
/// Postgres `Type`.
fn accepts(ty: &Type) -> bool
where
Self: Sized;
/// An adaptor method used internally by Rust-Postgres.
///
/// *All* implementations of this method should be generated by the
/// `to_sql_checked!()` macro.
fn to_sql_checked(
&self,
ty: &Type,
out: &mut BytesMut,
) -> Result<IsNull, Box<dyn Error + Sync + Send>>;
/// Specify the encode format
fn encode_format(&self, _ty: &Type) -> Format {
Format::Binary
}
}
/// Supported Postgres message format types
///
/// Using Text format in a message assumes a Postgres `SERVER_ENCODING` of `UTF8`
@@ -351,3 +424,52 @@ pub enum Format {
/// Compact, typed binary format
Binary,
}
impl ToSql for &str {
fn to_sql(&self, ty: &Type, w: &mut BytesMut) -> Result<IsNull, Box<dyn Error + Sync + Send>> {
match *ty {
ref ty if ty.name() == "ltree" => types::ltree_to_sql(self, w),
ref ty if ty.name() == "lquery" => types::lquery_to_sql(self, w),
ref ty if ty.name() == "ltxtquery" => types::ltxtquery_to_sql(self, w),
_ => types::text_to_sql(self, w),
}
Ok(IsNull::No)
}
fn accepts(ty: &Type) -> bool {
match *ty {
Type::VARCHAR | Type::TEXT | Type::BPCHAR | Type::NAME | Type::UNKNOWN => true,
ref ty
if (ty.name() == "citext"
|| ty.name() == "ltree"
|| ty.name() == "lquery"
|| ty.name() == "ltxtquery") =>
{
true
}
_ => false,
}
}
to_sql_checked!();
}
macro_rules! simple_to {
($t:ty, $f:ident, $($expected:ident),+) => {
impl ToSql for $t {
fn to_sql(&self,
_: &Type,
w: &mut BytesMut)
-> Result<IsNull, Box<dyn Error + Sync + Send>> {
types::$f(*self, w);
Ok(IsNull::No)
}
accepts!($($expected),+);
to_sql_checked!();
}
}
}
simple_to!(u32, oid_to_sql, OID);

View File

@@ -393,7 +393,7 @@ impl Inner {
}
}
pub const fn const_oid(&self) -> Oid {
pub fn oid(&self) -> Oid {
match *self {
Inner::Bool => 16,
Inner::Bytea => 17,
@@ -580,14 +580,7 @@ impl Inner {
Inner::TstzmultiRangeArray => 6153,
Inner::DatemultiRangeArray => 6155,
Inner::Int8multiRangeArray => 6157,
Inner::Other(_) => u32::MAX,
}
}
pub fn oid(&self) -> Oid {
match *self {
Inner::Other(ref u) => u.oid,
_ => self.const_oid(),
}
}
@@ -734,17 +727,17 @@ impl Inner {
Inner::JsonbArray => &Kind::Array(Type(Inner::Jsonb)),
Inner::AnyRange => &Kind::Pseudo,
Inner::EventTrigger => &Kind::Pseudo,
Inner::Int4Range => &const { Kind::Range(Inner::Int4.const_oid()) },
Inner::Int4Range => &Kind::Range(Type(Inner::Int4)),
Inner::Int4RangeArray => &Kind::Array(Type(Inner::Int4Range)),
Inner::NumRange => &const { Kind::Range(Inner::Numeric.const_oid()) },
Inner::NumRange => &Kind::Range(Type(Inner::Numeric)),
Inner::NumRangeArray => &Kind::Array(Type(Inner::NumRange)),
Inner::TsRange => &const { Kind::Range(Inner::Timestamp.const_oid()) },
Inner::TsRange => &Kind::Range(Type(Inner::Timestamp)),
Inner::TsRangeArray => &Kind::Array(Type(Inner::TsRange)),
Inner::TstzRange => &const { Kind::Range(Inner::Timestamptz.const_oid()) },
Inner::TstzRange => &Kind::Range(Type(Inner::Timestamptz)),
Inner::TstzRangeArray => &Kind::Array(Type(Inner::TstzRange)),
Inner::DateRange => &const { Kind::Range(Inner::Date.const_oid()) },
Inner::DateRange => &Kind::Range(Type(Inner::Date)),
Inner::DateRangeArray => &Kind::Array(Type(Inner::DateRange)),
Inner::Int8Range => &const { Kind::Range(Inner::Int8.const_oid()) },
Inner::Int8Range => &Kind::Range(Type(Inner::Int8)),
Inner::Int8RangeArray => &Kind::Array(Type(Inner::Int8Range)),
Inner::Jsonpath => &Kind::Simple,
Inner::JsonpathArray => &Kind::Array(Type(Inner::Jsonpath)),

View File

@@ -1,12 +1,14 @@
use std::collections::HashMap;
use std::fmt;
use std::net::IpAddr;
use std::sync::Arc;
use std::task::{Context, Poll};
use std::time::Duration;
use bytes::BytesMut;
use fallible_iterator::FallibleIterator;
use futures_util::{TryStreamExt, future, ready};
use parking_lot::Mutex;
use postgres_protocol2::message::backend::Message;
use postgres_protocol2::message::frontend;
use serde::{Deserialize, Serialize};
@@ -14,52 +16,29 @@ use tokio::sync::mpsc;
use crate::codec::{BackendMessages, FrontendMessage};
use crate::config::{Host, SslMode};
use crate::connection::{Request, RequestMessages};
use crate::query::RowStream;
use crate::simple_query::SimpleQueryStream;
use crate::types::{Oid, Type};
use crate::{
CancelToken, Error, ReadyForQueryStatus, SimpleQueryMessage, Transaction, TransactionBuilder,
query, simple_query,
CancelToken, Error, ReadyForQueryStatus, SimpleQueryMessage, Statement, Transaction,
TransactionBuilder, query, simple_query,
};
pub struct Responses {
/// new messages from conn
receiver: mpsc::Receiver<BackendMessages>,
/// current batch of messages
cur: BackendMessages,
/// number of total queries sent.
waiting: usize,
/// number of ReadyForQuery messages received.
received: usize,
}
impl Responses {
pub fn poll_next(&mut self, cx: &mut Context<'_>) -> Poll<Result<Message, Error>> {
loop {
// get the next saved message
if let Some(message) = self.cur.next().map_err(Error::parse)? {
let received = self.received;
// increase the query head if this is the last message.
if let Message::ReadyForQuery(_) = message {
self.received += 1;
}
// check if the client has skipped this query.
if received + 1 < self.waiting {
// grab the next message.
continue;
}
// convenience: turn the error messaage into a proper error.
let res = match message {
Message::ErrorResponse(body) => Err(Error::db(body)),
message => Ok(message),
};
return Poll::Ready(res);
match self.cur.next().map_err(Error::parse)? {
Some(Message::ErrorResponse(body)) => return Poll::Ready(Err(Error::db(body))),
Some(message) => return Poll::Ready(Ok(message)),
None => {}
}
// get the next batch of messages.
match ready!(self.receiver.poll_recv(cx)) {
Some(messages) => self.cur = messages,
None => return Poll::Ready(Err(Error::closed())),
@@ -76,87 +55,44 @@ impl Responses {
/// (corresponding to the queries in the [crate::prepare] module).
#[derive(Default)]
pub(crate) struct CachedTypeInfo {
/// A statement for basic information for a type from its
/// OID. Corresponds to [TYPEINFO_QUERY](crate::prepare::TYPEINFO_QUERY) (or its
/// fallback).
pub(crate) typeinfo: Option<Statement>,
/// Cache of types already looked up.
pub(crate) types: HashMap<Oid, Type>,
}
pub struct InnerClient {
sender: mpsc::UnboundedSender<FrontendMessage>,
responses: Responses,
sender: mpsc::UnboundedSender<Request>,
/// A buffer to use when writing out postgres commands.
buffer: BytesMut,
buffer: Mutex<BytesMut>,
}
impl InnerClient {
pub fn start(&mut self) -> Result<PartialQuery, Error> {
self.responses.waiting += 1;
Ok(PartialQuery(Some(self)))
pub fn send(&self, messages: RequestMessages) -> Result<Responses, Error> {
let (sender, receiver) = mpsc::channel(1);
let request = Request { messages, sender };
self.sender.send(request).map_err(|_| Error::closed())?;
Ok(Responses {
receiver,
cur: BackendMessages::empty(),
})
}
// pub fn send_with_sync<F>(&mut self, f: F) -> Result<&mut Responses, Error>
// where
// F: FnOnce(&mut BytesMut) -> Result<(), Error>,
// {
// self.start()?.send_with_sync(f)
// }
pub fn send_simple_query(&mut self, query: &str) -> Result<&mut Responses, Error> {
self.responses.waiting += 1;
self.buffer.clear();
// simple queries do not need sync.
frontend::query(query, &mut self.buffer).map_err(Error::encode)?;
let buf = self.buffer.split().freeze();
self.send_message(FrontendMessage::Raw(buf))
}
fn send_message(&mut self, messages: FrontendMessage) -> Result<&mut Responses, Error> {
self.sender.send(messages).map_err(|_| Error::closed())?;
Ok(&mut self.responses)
}
}
pub struct PartialQuery<'a>(Option<&'a mut InnerClient>);
impl Drop for PartialQuery<'_> {
fn drop(&mut self) {
if let Some(client) = self.0.take() {
client.buffer.clear();
frontend::sync(&mut client.buffer);
let buf = client.buffer.split().freeze();
let _ = client.send_message(FrontendMessage::Raw(buf));
}
}
}
impl<'a> PartialQuery<'a> {
pub fn send_with_flush<F>(&mut self, f: F) -> Result<&mut Responses, Error>
/// Call the given function with a buffer to be used when writing out
/// postgres commands.
pub fn with_buf<F, R>(&self, f: F) -> R
where
F: FnOnce(&mut BytesMut) -> Result<(), Error>,
F: FnOnce(&mut BytesMut) -> R,
{
let client = self.0.as_deref_mut().unwrap();
client.buffer.clear();
f(&mut client.buffer)?;
frontend::flush(&mut client.buffer);
let buf = client.buffer.split().freeze();
client.send_message(FrontendMessage::Raw(buf))
}
pub fn send_with_sync<F>(mut self, f: F) -> Result<&'a mut Responses, Error>
where
F: FnOnce(&mut BytesMut) -> Result<(), Error>,
{
let client = self.0.as_deref_mut().unwrap();
client.buffer.clear();
f(&mut client.buffer)?;
frontend::sync(&mut client.buffer);
let buf = client.buffer.split().freeze();
let _ = client.send_message(FrontendMessage::Raw(buf));
Ok(&mut self.0.take().unwrap().responses)
let mut buffer = self.buffer.lock();
let r = f(&mut buffer);
buffer.clear();
r
}
}
@@ -173,7 +109,7 @@ pub struct SocketConfig {
/// The client is one half of what is returned when a connection is established. Users interact with the database
/// through this client object.
pub struct Client {
inner: InnerClient,
inner: Arc<InnerClient>,
cached_typeinfo: CachedTypeInfo,
socket_config: SocketConfig,
@@ -184,24 +120,17 @@ pub struct Client {
impl Client {
pub(crate) fn new(
sender: mpsc::UnboundedSender<FrontendMessage>,
receiver: mpsc::Receiver<BackendMessages>,
sender: mpsc::UnboundedSender<Request>,
socket_config: SocketConfig,
ssl_mode: SslMode,
process_id: i32,
secret_key: i32,
) -> Client {
Client {
inner: InnerClient {
inner: Arc::new(InnerClient {
sender,
responses: Responses {
receiver,
cur: BackendMessages::empty(),
waiting: 0,
received: 0,
},
buffer: Default::default(),
},
}),
cached_typeinfo: Default::default(),
socket_config,
@@ -216,29 +145,19 @@ impl Client {
self.process_id
}
pub(crate) fn inner_mut(&mut self) -> &mut InnerClient {
&mut self.inner
pub(crate) fn inner(&self) -> &Arc<InnerClient> {
&self.inner
}
/// Pass text directly to the Postgres backend to allow it to sort out typing itself and
/// to save a roundtrip
pub async fn query_raw_txt<S, I>(
&mut self,
statement: &str,
params: I,
) -> Result<RowStream, Error>
pub async fn query_raw_txt<S, I>(&self, statement: &str, params: I) -> Result<RowStream, Error>
where
S: AsRef<str>,
I: IntoIterator<Item = Option<S>>,
I::IntoIter: ExactSizeIterator,
{
query::query_txt(
&mut self.inner,
&mut self.cached_typeinfo,
statement,
params,
)
.await
query::query_txt(&self.inner, statement, params).await
}
/// Executes a sequence of SQL statements using the simple query protocol, returning the resulting rows.
@@ -254,15 +173,12 @@ impl Client {
/// Prepared statements should be use for any query which contains user-specified data, as they provided the
/// functionality to safely embed that data in the request. Do not form statements via string concatenation and pass
/// them to this method!
pub async fn simple_query(&mut self, query: &str) -> Result<Vec<SimpleQueryMessage>, Error> {
pub async fn simple_query(&self, query: &str) -> Result<Vec<SimpleQueryMessage>, Error> {
self.simple_query_raw(query).await?.try_collect().await
}
pub(crate) async fn simple_query_raw(
&mut self,
query: &str,
) -> Result<SimpleQueryStream, Error> {
simple_query::simple_query(self.inner_mut(), query).await
pub(crate) async fn simple_query_raw(&self, query: &str) -> Result<SimpleQueryStream, Error> {
simple_query::simple_query(self.inner(), query).await
}
/// Executes a sequence of SQL statements using the simple query protocol.
@@ -275,11 +191,15 @@ impl Client {
/// Prepared statements should be use for any query which contains user-specified data, as they provided the
/// functionality to safely embed that data in the request. Do not form statements via string concatenation and pass
/// them to this method!
pub async fn batch_execute(&mut self, query: &str) -> Result<ReadyForQueryStatus, Error> {
simple_query::batch_execute(self.inner_mut(), query).await
pub async fn batch_execute(&self, query: &str) -> Result<ReadyForQueryStatus, Error> {
simple_query::batch_execute(self.inner(), query).await
}
pub async fn discard_all(&mut self) -> Result<ReadyForQueryStatus, Error> {
// clear the prepared statements that are about to be nuked from the postgres session
self.cached_typeinfo.typeinfo = None;
self.batch_execute("discard all").await
}
@@ -288,7 +208,7 @@ impl Client {
/// The transaction will roll back by default - use the `commit` method to commit it.
pub async fn transaction(&mut self) -> Result<Transaction<'_>, Error> {
struct RollbackIfNotDone<'me> {
client: &'me mut Client,
client: &'me Client,
done: bool,
}
@@ -298,7 +218,14 @@ impl Client {
return;
}
let _ = self.client.inner.send_simple_query("ROLLBACK");
let buf = self.client.inner().with_buf(|buf| {
frontend::query("ROLLBACK", buf).unwrap();
buf.split().freeze()
});
let _ = self
.client
.inner()
.send(RequestMessages::Single(FrontendMessage::Raw(buf)));
}
}
@@ -312,7 +239,7 @@ impl Client {
client: self,
done: false,
};
cleaner.client.batch_execute("BEGIN").await?;
self.batch_execute("BEGIN").await?;
cleaner.done = true;
}
@@ -338,6 +265,11 @@ impl Client {
}
}
/// Query for type information
pub(crate) async fn get_type_inner(&mut self, oid: Oid) -> Result<Type, Error> {
crate::prepare::get_type(&self.inner, &mut self.cached_typeinfo, oid).await
}
/// Determines if the connection to the server has already closed.
///
/// In that case, all future queries will fail.

View File

@@ -1,16 +1,21 @@
use std::io;
use bytes::{Bytes, BytesMut};
use bytes::{Buf, Bytes, BytesMut};
use fallible_iterator::FallibleIterator;
use postgres_protocol2::message::backend;
use postgres_protocol2::message::frontend::CopyData;
use tokio_util::codec::{Decoder, Encoder};
pub enum FrontendMessage {
Raw(Bytes),
CopyData(CopyData<Box<dyn Buf + Send>>),
}
pub enum BackendMessage {
Normal { messages: BackendMessages },
Normal {
messages: BackendMessages,
request_complete: bool,
},
Async(backend::Message),
}
@@ -39,6 +44,7 @@ impl Encoder<FrontendMessage> for PostgresCodec {
fn encode(&mut self, item: FrontendMessage, dst: &mut BytesMut) -> io::Result<()> {
match item {
FrontendMessage::Raw(buf) => dst.extend_from_slice(&buf),
FrontendMessage::CopyData(data) => data.write(dst),
}
Ok(())
@@ -51,6 +57,7 @@ impl Decoder for PostgresCodec {
fn decode(&mut self, src: &mut BytesMut) -> Result<Option<BackendMessage>, io::Error> {
let mut idx = 0;
let mut request_complete = false;
while let Some(header) = backend::Header::parse(&src[idx..])? {
let len = header.len() as usize + 1;
@@ -75,6 +82,7 @@ impl Decoder for PostgresCodec {
idx += len;
if header.tag() == backend::READY_FOR_QUERY_TAG {
request_complete = true;
break;
}
}
@@ -84,6 +92,7 @@ impl Decoder for PostgresCodec {
} else {
Ok(Some(BackendMessage::Normal {
messages: BackendMessages(src.split_to(idx)),
request_complete,
}))
}
}

View File

@@ -59,11 +59,9 @@ where
connect_timeout: config.connect_timeout,
};
let (client_tx, conn_rx) = mpsc::unbounded_channel();
let (conn_tx, client_rx) = mpsc::channel(4);
let (sender, receiver) = mpsc::unbounded_channel();
let client = Client::new(
client_tx,
client_rx,
sender,
socket_config,
config.ssl_mode,
process_id,
@@ -76,7 +74,7 @@ where
.map(|m| BackendMessage::Async(Message::NoticeResponse(m)))
.collect();
let connection = Connection::new(stream, delayed, parameters, conn_tx, conn_rx);
let connection = Connection::new(stream, delayed, parameters, receiver);
Ok((client, connection))
}

View File

@@ -4,6 +4,7 @@ use std::pin::Pin;
use std::task::{Context, Poll};
use bytes::BytesMut;
use fallible_iterator::FallibleIterator;
use futures_util::{Sink, Stream, ready};
use postgres_protocol2::message::backend::Message;
use postgres_protocol2::message::frontend;
@@ -18,12 +19,30 @@ use crate::error::DbError;
use crate::maybe_tls_stream::MaybeTlsStream;
use crate::{AsyncMessage, Error, Notification};
pub enum RequestMessages {
Single(FrontendMessage),
}
pub struct Request {
pub messages: RequestMessages,
pub sender: mpsc::Sender<BackendMessages>,
}
pub struct Response {
sender: PollSender<BackendMessages>,
}
#[derive(PartialEq, Debug)]
enum State {
Active,
Closing,
}
enum WriteReady {
Terminating,
WaitingOnRead,
}
/// A connection to a PostgreSQL database.
///
/// This is one half of what is returned when a new connection is established. It performs the actual IO with the
@@ -37,11 +56,9 @@ pub struct Connection<S, T> {
pub stream: Framed<MaybeTlsStream<S, T>, PostgresCodec>,
/// HACK: we need this in the Neon Proxy to forward params.
pub parameters: HashMap<String, String>,
sender: PollSender<BackendMessages>,
receiver: mpsc::UnboundedReceiver<FrontendMessage>,
receiver: mpsc::UnboundedReceiver<Request>,
pending_responses: VecDeque<BackendMessage>,
responses: VecDeque<Response>,
state: State,
}
@@ -54,15 +71,14 @@ where
stream: Framed<MaybeTlsStream<S, T>, PostgresCodec>,
pending_responses: VecDeque<BackendMessage>,
parameters: HashMap<String, String>,
sender: mpsc::Sender<BackendMessages>,
receiver: mpsc::UnboundedReceiver<FrontendMessage>,
receiver: mpsc::UnboundedReceiver<Request>,
) -> Connection<S, T> {
Connection {
stream,
parameters,
sender: PollSender::new(sender),
receiver,
pending_responses,
responses: VecDeque::new(),
state: State::Active,
}
}
@@ -94,7 +110,7 @@ where
}
};
let messages = match message {
let (mut messages, request_complete) = match message {
BackendMessage::Async(Message::NoticeResponse(body)) => {
let error = DbError::parse(&mut body.fields()).map_err(Error::parse)?;
return Poll::Ready(Ok(AsyncMessage::Notice(error)));
@@ -115,19 +131,41 @@ where
continue;
}
BackendMessage::Async(_) => unreachable!(),
BackendMessage::Normal { messages } => messages,
BackendMessage::Normal {
messages,
request_complete,
} => (messages, request_complete),
};
match self.sender.poll_reserve(cx) {
let mut response = match self.responses.pop_front() {
Some(response) => response,
None => match messages.next().map_err(Error::parse)? {
Some(Message::ErrorResponse(error)) => {
return Poll::Ready(Err(Error::db(error)));
}
_ => return Poll::Ready(Err(Error::unexpected_message())),
},
};
match response.sender.poll_reserve(cx) {
Poll::Ready(Ok(())) => {
let _ = self.sender.send_item(messages);
let _ = response.sender.send_item(messages);
if !request_complete {
self.responses.push_front(response);
}
}
Poll::Ready(Err(_)) => {
return Poll::Ready(Err(Error::closed()));
// we need to keep paging through the rest of the messages even if the receiver's hung up
if !request_complete {
self.responses.push_front(response);
}
}
Poll::Pending => {
self.pending_responses
.push_back(BackendMessage::Normal { messages });
self.responses.push_front(response);
self.pending_responses.push_back(BackendMessage::Normal {
messages,
request_complete,
});
trace!("poll_read: waiting on sender");
return Poll::Pending;
}
@@ -136,7 +174,7 @@ where
}
/// Fetch the next client request and enqueue the response sender.
fn poll_request(&mut self, cx: &mut Context<'_>) -> Poll<Option<FrontendMessage>> {
fn poll_request(&mut self, cx: &mut Context<'_>) -> Poll<Option<RequestMessages>> {
if self.receiver.is_closed() {
return Poll::Ready(None);
}
@@ -144,7 +182,10 @@ where
match self.receiver.poll_recv(cx) {
Poll::Ready(Some(request)) => {
trace!("polled new request");
Poll::Ready(Some(request))
self.responses.push_back(Response {
sender: PollSender::new(request.sender),
});
Poll::Ready(Some(request.messages))
}
Poll::Ready(None) => Poll::Ready(None),
Poll::Pending => Poll::Pending,
@@ -153,7 +194,7 @@ where
/// Process client requests and write them to the postgres connection, flushing if necessary.
/// client -> postgres
fn poll_write(&mut self, cx: &mut Context<'_>) -> Poll<Result<(), Error>> {
fn poll_write(&mut self, cx: &mut Context<'_>) -> Poll<Result<WriteReady, Error>> {
loop {
if Pin::new(&mut self.stream)
.poll_ready(cx)
@@ -168,14 +209,14 @@ where
match self.poll_request(cx) {
// send the message to postgres
Poll::Ready(Some(request)) => {
Poll::Ready(Some(RequestMessages::Single(request))) => {
Pin::new(&mut self.stream)
.start_send(request)
.map_err(Error::io)?;
}
// No more messages from the client, and no more responses to wait for.
// Send a terminate message to postgres
Poll::Ready(None) => {
Poll::Ready(None) if self.responses.is_empty() => {
trace!("poll_write: at eof, terminating");
let mut request = BytesMut::new();
frontend::terminate(&mut request);
@@ -187,7 +228,16 @@ where
trace!("poll_write: sent eof, closing");
trace!("poll_write: done");
return Poll::Ready(Ok(()));
return Poll::Ready(Ok(WriteReady::Terminating));
}
// No more messages from the client, but there are still some responses to wait for.
Poll::Ready(None) => {
trace!(
"poll_write: at eof, pending responses {}",
self.responses.len()
);
ready!(self.poll_flush(cx))?;
return Poll::Ready(Ok(WriteReady::WaitingOnRead));
}
// Still waiting for a message from the client.
Poll::Pending => {
@@ -248,7 +298,7 @@ where
// if the state is still active, try read from and write to postgres.
let message = self.poll_read(cx)?;
let closing = self.poll_write(cx)?;
if let Poll::Ready(()) = closing {
if let Poll::Ready(WriteReady::Terminating) = closing {
self.state = State::Closing;
}

View File

@@ -86,27 +86,6 @@ pub struct DbError {
}
impl DbError {
pub fn new_test_error(code: SqlState, message: String) -> Self {
DbError {
severity: "ERROR".to_string(),
parsed_severity: Some(Severity::Error),
code,
message,
detail: None,
hint: None,
position: None,
where_: None,
schema: None,
table: None,
column: None,
datatype: None,
constraint: None,
file: None,
line: None,
routine: None,
}
}
pub(crate) fn parse(fields: &mut ErrorFields<'_>) -> io::Result<DbError> {
let mut severity = None;
let mut parsed_severity = None;

View File

@@ -1,6 +1,9 @@
#![allow(async_fn_in_trait)]
use postgres_protocol2::Oid;
use crate::query::RowStream;
use crate::types::Type;
use crate::{Client, Error, Transaction};
mod private {
@@ -12,17 +15,20 @@ mod private {
/// This trait is "sealed", and cannot be implemented outside of this crate.
pub trait GenericClient: private::Sealed {
/// Like `Client::query_raw_txt`.
async fn query_raw_txt<S, I>(&mut self, statement: &str, params: I) -> Result<RowStream, Error>
async fn query_raw_txt<S, I>(&self, statement: &str, params: I) -> Result<RowStream, Error>
where
S: AsRef<str> + Sync + Send,
I: IntoIterator<Item = Option<S>> + Sync + Send,
I::IntoIter: ExactSizeIterator + Sync + Send;
/// Query for type information
async fn get_type(&mut self, oid: Oid) -> Result<Type, Error>;
}
impl private::Sealed for Client {}
impl GenericClient for Client {
async fn query_raw_txt<S, I>(&mut self, statement: &str, params: I) -> Result<RowStream, Error>
async fn query_raw_txt<S, I>(&self, statement: &str, params: I) -> Result<RowStream, Error>
where
S: AsRef<str> + Sync + Send,
I: IntoIterator<Item = Option<S>> + Sync + Send,
@@ -30,12 +36,17 @@ impl GenericClient for Client {
{
self.query_raw_txt(statement, params).await
}
/// Query for type information
async fn get_type(&mut self, oid: Oid) -> Result<Type, Error> {
self.get_type_inner(oid).await
}
}
impl private::Sealed for Transaction<'_> {}
impl GenericClient for Transaction<'_> {
async fn query_raw_txt<S, I>(&mut self, statement: &str, params: I) -> Result<RowStream, Error>
async fn query_raw_txt<S, I>(&self, statement: &str, params: I) -> Result<RowStream, Error>
where
S: AsRef<str> + Sync + Send,
I: IntoIterator<Item = Option<S>> + Sync + Send,
@@ -43,4 +54,9 @@ impl GenericClient for Transaction<'_> {
{
self.query_raw_txt(statement, params).await
}
/// Query for type information
async fn get_type(&mut self, oid: Oid) -> Result<Type, Error> {
self.client_mut().get_type(oid).await
}
}

View File

@@ -18,6 +18,7 @@ pub use crate::statement::{Column, Statement};
pub use crate::tls::NoTls;
pub use crate::transaction::Transaction;
pub use crate::transaction_builder::{IsolationLevel, TransactionBuilder};
use crate::types::ToSql;
/// After executing a query, the connection will be in one of these states
#[derive(Clone, Copy, Debug, PartialEq)]
@@ -119,3 +120,9 @@ pub enum SimpleQueryMessage {
/// The number of rows modified or selected is returned.
CommandComplete(u64),
}
fn slice_iter<'a>(
s: &'a [&'a (dyn ToSql + Sync)],
) -> impl ExactSizeIterator<Item = &'a (dyn ToSql + Sync)> + 'a {
s.iter().map(|s| *s as _)
}

View File

@@ -1,14 +1,19 @@
use bytes::BytesMut;
use fallible_iterator::FallibleIterator;
use postgres_protocol2::IsNull;
use postgres_protocol2::message::backend::{Message, RowDescriptionBody};
use postgres_protocol2::message::frontend;
use postgres_protocol2::types::oid_to_sql;
use postgres_types2::Format;
use std::future::Future;
use std::pin::Pin;
use std::sync::Arc;
use crate::client::{CachedTypeInfo, PartialQuery, Responses};
use bytes::Bytes;
use fallible_iterator::FallibleIterator;
use futures_util::{TryStreamExt, pin_mut};
use postgres_protocol2::message::backend::Message;
use postgres_protocol2::message::frontend;
use tracing::debug;
use crate::client::{CachedTypeInfo, InnerClient};
use crate::codec::FrontendMessage;
use crate::connection::RequestMessages;
use crate::types::{Kind, Oid, Type};
use crate::{Column, Error, Row, Statement};
use crate::{Column, Error, Statement, query, slice_iter};
pub(crate) const TYPEINFO_QUERY: &str = "\
SELECT t.typname, t.typtype, t.typelem, r.rngsubtype, t.typbasetype, n.nspname, t.typrelid
@@ -18,51 +23,22 @@ INNER JOIN pg_catalog.pg_namespace n ON t.typnamespace = n.oid
WHERE t.oid = $1
";
/// we need to make sure we close this prepared statement.
struct CloseStmt<'a, 'b> {
client: Option<&'a mut PartialQuery<'b>>,
name: &'static str,
}
impl<'a> CloseStmt<'a, '_> {
fn close(mut self) -> Result<&'a mut Responses, Error> {
let client = self.client.take().unwrap();
client.send_with_flush(|buf| {
frontend::close(b'S', self.name, buf).map_err(Error::encode)?;
Ok(())
})
}
}
impl Drop for CloseStmt<'_, '_> {
fn drop(&mut self) {
if let Some(client) = self.client.take() {
let _ = client.send_with_flush(|buf| {
frontend::close(b'S', self.name, buf).map_err(Error::encode)?;
Ok(())
});
}
}
}
async fn prepare_typecheck(
client: &mut PartialQuery<'_>,
client: &Arc<InnerClient>,
name: &'static str,
query: &str,
types: &[Type],
) -> Result<Statement, Error> {
let responses = client.send_with_flush(|buf| {
frontend::parse(name, query, [], buf).map_err(Error::encode)?;
frontend::describe(b'S', name, buf).map_err(Error::encode)?;
Ok(())
})?;
let buf = encode(client, name, query, types)?;
let mut responses = client.send(RequestMessages::Single(FrontendMessage::Raw(buf)))?;
match responses.next().await? {
Message::ParseComplete => {}
_ => return Err(Error::unexpected_message()),
}
match responses.next().await? {
Message::ParameterDescription(_) => {}
let parameter_description = match responses.next().await? {
Message::ParameterDescription(body) => body,
_ => return Err(Error::unexpected_message()),
};
@@ -72,6 +48,13 @@ async fn prepare_typecheck(
_ => return Err(Error::unexpected_message()),
};
let mut parameters = vec![];
let mut it = parameter_description.parameters();
while let Some(oid) = it.next().map_err(Error::parse)? {
let type_ = Type::from_oid(oid).ok_or_else(Error::unexpected_message)?;
parameters.push(type_);
}
let mut columns = vec![];
if let Some(row_description) = row_description {
let mut it = row_description.fields();
@@ -82,168 +65,98 @@ async fn prepare_typecheck(
}
}
Ok(Statement::new(name, columns))
Ok(Statement::new(client, name, parameters, columns))
}
fn try_from_cache(typecache: &CachedTypeInfo, oid: Oid) -> Option<Type> {
fn encode(client: &InnerClient, name: &str, query: &str, types: &[Type]) -> Result<Bytes, Error> {
if types.is_empty() {
debug!("preparing query {}: {}", name, query);
} else {
debug!("preparing query {} with types {:?}: {}", name, types, query);
}
client.with_buf(|buf| {
frontend::parse(name, query, types.iter().map(Type::oid), buf).map_err(Error::encode)?;
frontend::describe(b'S', name, buf).map_err(Error::encode)?;
frontend::sync(buf);
Ok(buf.split().freeze())
})
}
pub async fn get_type(
client: &Arc<InnerClient>,
typecache: &mut CachedTypeInfo,
oid: Oid,
) -> Result<Type, Error> {
if let Some(type_) = Type::from_oid(oid) {
return Some(type_);
return Ok(type_);
}
if let Some(type_) = typecache.types.get(&oid) {
return Some(type_.clone());
return Ok(type_.clone());
};
None
}
let stmt = typeinfo_statement(client, typecache).await?;
pub async fn parse_row_description(
client: &mut PartialQuery<'_>,
typecache: &mut CachedTypeInfo,
row_description: Option<RowDescriptionBody>,
) -> Result<Vec<Column>, Error> {
let mut columns = vec![];
let rows = query::query(client, stmt, slice_iter(&[&oid])).await?;
pin_mut!(rows);
if let Some(row_description) = row_description {
let mut it = row_description.fields();
while let Some(field) = it.next().map_err(Error::parse)? {
let type_ = try_from_cache(typecache, field.type_oid()).unwrap_or(Type::UNKNOWN);
let column = Column::new(field.name().to_string(), type_, field);
columns.push(column);
}
}
let all_known = columns.iter().all(|c| c.type_ != Type::UNKNOWN);
if all_known {
// all known, return early.
return Ok(columns);
}
let typeinfo = "neon_proxy_typeinfo";
// make sure to close the typeinfo statement before exiting.
let mut guard = CloseStmt {
name: typeinfo,
client: None,
};
let client = guard.client.insert(client);
// get the typeinfo statement.
let stmt = prepare_typecheck(client, typeinfo, TYPEINFO_QUERY).await?;
for column in &mut columns {
column.type_ = get_type(client, typecache, &stmt, column.type_oid()).await?;
}
// cancel the close guard.
let responses = guard.close()?;
match responses.next().await? {
Message::CloseComplete => {}
_ => return Err(Error::unexpected_message()),
}
Ok(columns)
}
async fn get_type(
client: &mut PartialQuery<'_>,
typecache: &mut CachedTypeInfo,
stmt: &Statement,
mut oid: Oid,
) -> Result<Type, Error> {
let mut stack = vec![];
let mut type_ = loop {
if let Some(type_) = try_from_cache(typecache, oid) {
break type_;
}
let row = exec(client, stmt, oid).await?;
if stack.len() > 8 {
return Err(Error::unexpected_message());
}
let name: String = row.try_get(0)?;
let type_: i8 = row.try_get(1)?;
let elem_oid: Oid = row.try_get(2)?;
let rngsubtype: Option<Oid> = row.try_get(3)?;
let basetype: Oid = row.try_get(4)?;
let schema: String = row.try_get(5)?;
let relid: Oid = row.try_get(6)?;
let kind = if type_ == b'e' as i8 {
Kind::Enum
} else if type_ == b'p' as i8 {
Kind::Pseudo
} else if basetype != 0 {
Kind::Domain(basetype)
} else if elem_oid != 0 {
stack.push((name, oid, schema));
oid = elem_oid;
continue;
} else if relid != 0 {
Kind::Composite(relid)
} else if let Some(rngsubtype) = rngsubtype {
Kind::Range(rngsubtype)
} else {
Kind::Simple
};
let type_ = Type::new(name, oid, kind, schema);
typecache.types.insert(oid, type_.clone());
break type_;
let row = match rows.try_next().await? {
Some(row) => row,
None => return Err(Error::unexpected_message()),
};
while let Some((name, oid, schema)) = stack.pop() {
type_ = Type::new(name, oid, Kind::Array(type_), schema);
typecache.types.insert(oid, type_.clone());
}
let name: String = row.try_get(0)?;
let type_: i8 = row.try_get(1)?;
let elem_oid: Oid = row.try_get(2)?;
let rngsubtype: Option<Oid> = row.try_get(3)?;
let basetype: Oid = row.try_get(4)?;
let schema: String = row.try_get(5)?;
let relid: Oid = row.try_get(6)?;
let kind = if type_ == b'e' as i8 {
Kind::Enum
} else if type_ == b'p' as i8 {
Kind::Pseudo
} else if basetype != 0 {
Kind::Domain(basetype)
} else if elem_oid != 0 {
let type_ = get_type_rec(client, typecache, elem_oid).await?;
Kind::Array(type_)
} else if relid != 0 {
Kind::Composite(relid)
} else if let Some(rngsubtype) = rngsubtype {
let type_ = get_type_rec(client, typecache, rngsubtype).await?;
Kind::Range(type_)
} else {
Kind::Simple
};
let type_ = Type::new(name, oid, kind, schema);
typecache.types.insert(oid, type_.clone());
Ok(type_)
}
/// exec the typeinfo statement returning one row.
async fn exec(
client: &mut PartialQuery<'_>,
statement: &Statement,
param: Oid,
) -> Result<Row, Error> {
let responses = client.send_with_flush(|buf| {
encode_bind(statement, param, "", buf);
frontend::execute("", 0, buf).map_err(Error::encode)?;
Ok(())
})?;
fn get_type_rec<'a>(
client: &'a Arc<InnerClient>,
typecache: &'a mut CachedTypeInfo,
oid: Oid,
) -> Pin<Box<dyn Future<Output = Result<Type, Error>> + Send + 'a>> {
Box::pin(get_type(client, typecache, oid))
}
match responses.next().await? {
Message::BindComplete => {}
_ => return Err(Error::unexpected_message()),
async fn typeinfo_statement(
client: &Arc<InnerClient>,
typecache: &mut CachedTypeInfo,
) -> Result<Statement, Error> {
if let Some(stmt) = &typecache.typeinfo {
return Ok(stmt.clone());
}
let row = match responses.next().await? {
Message::DataRow(body) => Row::new(statement.clone(), body, Format::Binary)?,
_ => return Err(Error::unexpected_message()),
};
let typeinfo = "neon_proxy_typeinfo";
let stmt = prepare_typecheck(client, typeinfo, TYPEINFO_QUERY, &[]).await?;
match responses.next().await? {
Message::CommandComplete(_) => {}
_ => return Err(Error::unexpected_message()),
};
Ok(row)
}
fn encode_bind(statement: &Statement, param: Oid, portal: &str, buf: &mut BytesMut) {
frontend::bind(
portal,
statement.name(),
[Format::Binary as i16],
[param],
|param, buf| {
oid_to_sql(param, buf);
Ok(IsNull::No)
},
[Format::Binary as i16],
buf,
)
.unwrap();
typecache.typeinfo = Some(stmt.clone());
Ok(stmt)
}

View File

@@ -1,43 +1,76 @@
use std::fmt;
use std::marker::PhantomPinned;
use std::pin::Pin;
use std::sync::Arc;
use std::task::{Context, Poll};
use bytes::BufMut;
use bytes::{BufMut, Bytes, BytesMut};
use fallible_iterator::FallibleIterator;
use futures_util::{Stream, ready};
use pin_project_lite::pin_project;
use postgres_protocol2::message::backend::Message;
use postgres_protocol2::message::frontend;
use postgres_types2::Format;
use postgres_types2::{Format, ToSql, Type};
use tracing::debug;
use crate::client::{CachedTypeInfo, InnerClient, Responses};
use crate::{Error, ReadyForQueryStatus, Row, Statement};
use crate::client::{InnerClient, Responses};
use crate::codec::FrontendMessage;
use crate::connection::RequestMessages;
use crate::types::IsNull;
use crate::{Column, Error, ReadyForQueryStatus, Row, Statement};
pub async fn query_txt<'a, S, I>(
client: &'a mut InnerClient,
typecache: &mut CachedTypeInfo,
struct BorrowToSqlParamsDebug<'a>(&'a [&'a (dyn ToSql + Sync)]);
impl fmt::Debug for BorrowToSqlParamsDebug<'_> {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
f.debug_list().entries(self.0.iter()).finish()
}
}
pub async fn query<'a, I>(
client: &InnerClient,
statement: Statement,
params: I,
) -> Result<RowStream, Error>
where
I: IntoIterator<Item = &'a (dyn ToSql + Sync)>,
I::IntoIter: ExactSizeIterator,
{
let buf = if tracing::enabled!(tracing::Level::DEBUG) {
let params = params.into_iter().collect::<Vec<_>>();
debug!(
"executing statement {} with parameters: {:?}",
statement.name(),
BorrowToSqlParamsDebug(params.as_slice()),
);
encode(client, &statement, params)?
} else {
encode(client, &statement, params)?
};
let responses = start(client, buf).await?;
Ok(RowStream {
statement,
responses,
command_tag: None,
status: ReadyForQueryStatus::Unknown,
output_format: Format::Binary,
_p: PhantomPinned,
})
}
pub async fn query_txt<S, I>(
client: &Arc<InnerClient>,
query: &str,
params: I,
) -> Result<RowStream<'a>, Error>
) -> Result<RowStream, Error>
where
S: AsRef<str>,
I: IntoIterator<Item = Option<S>>,
I::IntoIter: ExactSizeIterator,
{
let params = params.into_iter();
let mut client = client.start()?;
// Flow:
// 1. Parse the query
// 2. Inspect the row description for OIDs
// 3. If there's any OIDs we don't already know about, perform the typeinfo routine
// 4. Execute the query
// 5. Sync.
//
// The typeinfo routine:
// 1. Parse the typeinfo query
// 2. Execute the query on each OID
// 3. If the result does not match an OID we know, repeat 2.
// parse the query and get type info
let responses = client.send_with_flush(|buf| {
let buf = client.with_buf(|buf| {
frontend::parse(
"", // unnamed prepared statement
query, // query to parse
@@ -46,30 +79,7 @@ where
)
.map_err(Error::encode)?;
frontend::describe(b'S', "", buf).map_err(Error::encode)?;
Ok(())
})?;
match responses.next().await? {
Message::ParseComplete => {}
_ => return Err(Error::unexpected_message()),
}
match responses.next().await? {
Message::ParameterDescription(_) => {}
_ => return Err(Error::unexpected_message()),
};
let row_description = match responses.next().await? {
Message::RowDescription(body) => Some(body),
Message::NoData => None,
_ => return Err(Error::unexpected_message()),
};
let columns =
crate::prepare::parse_row_description(&mut client, typecache, row_description).await?;
let responses = client.send_with_sync(|buf| {
// Bind, pass params as text, retrieve as text
// Bind, pass params as text, retrieve as binary
match frontend::bind(
"", // empty string selects the unnamed portal
"", // unnamed prepared statement
@@ -92,55 +102,173 @@ where
// Execute
frontend::execute("", 0, buf).map_err(Error::encode)?;
// Sync
frontend::sync(buf);
Ok(())
Ok(buf.split().freeze())
})?;
// now read the responses
let mut responses = client.send(RequestMessages::Single(FrontendMessage::Raw(buf)))?;
match responses.next().await? {
Message::ParseComplete => {}
_ => return Err(Error::unexpected_message()),
}
let parameter_description = match responses.next().await? {
Message::ParameterDescription(body) => body,
_ => return Err(Error::unexpected_message()),
};
let row_description = match responses.next().await? {
Message::RowDescription(body) => Some(body),
Message::NoData => None,
_ => return Err(Error::unexpected_message()),
};
match responses.next().await? {
Message::BindComplete => {}
_ => return Err(Error::unexpected_message()),
}
let mut parameters = vec![];
let mut it = parameter_description.parameters();
while let Some(oid) = it.next().map_err(Error::parse)? {
let type_ = Type::from_oid(oid).unwrap_or(Type::UNKNOWN);
parameters.push(type_);
}
let mut columns = vec![];
if let Some(row_description) = row_description {
let mut it = row_description.fields();
while let Some(field) = it.next().map_err(Error::parse)? {
let type_ = Type::from_oid(field.type_oid()).unwrap_or(Type::UNKNOWN);
let column = Column::new(field.name().to_string(), type_, field);
columns.push(column);
}
}
Ok(RowStream {
statement: Statement::new_anonymous(parameters, columns),
responses,
statement: Statement::new("", columns),
command_tag: None,
status: ReadyForQueryStatus::Unknown,
output_format: Format::Text,
_p: PhantomPinned,
})
}
/// A stream of table rows.
pub struct RowStream<'a> {
responses: &'a mut Responses,
output_format: Format,
pub statement: Statement,
pub command_tag: Option<String>,
pub status: ReadyForQueryStatus,
async fn start(client: &InnerClient, buf: Bytes) -> Result<Responses, Error> {
let mut responses = client.send(RequestMessages::Single(FrontendMessage::Raw(buf)))?;
match responses.next().await? {
Message::BindComplete => {}
_ => return Err(Error::unexpected_message()),
}
Ok(responses)
}
impl Stream for RowStream<'_> {
pub fn encode<'a, I>(client: &InnerClient, statement: &Statement, params: I) -> Result<Bytes, Error>
where
I: IntoIterator<Item = &'a (dyn ToSql + Sync)>,
I::IntoIter: ExactSizeIterator,
{
client.with_buf(|buf| {
encode_bind(statement, params, "", buf)?;
frontend::execute("", 0, buf).map_err(Error::encode)?;
frontend::sync(buf);
Ok(buf.split().freeze())
})
}
pub fn encode_bind<'a, I>(
statement: &Statement,
params: I,
portal: &str,
buf: &mut BytesMut,
) -> Result<(), Error>
where
I: IntoIterator<Item = &'a (dyn ToSql + Sync)>,
I::IntoIter: ExactSizeIterator,
{
let param_types = statement.params();
let params = params.into_iter();
assert!(
param_types.len() == params.len(),
"expected {} parameters but got {}",
param_types.len(),
params.len()
);
let (param_formats, params): (Vec<_>, Vec<_>) = params
.zip(param_types.iter())
.map(|(p, ty)| (p.encode_format(ty) as i16, p))
.unzip();
let params = params.into_iter();
let mut error_idx = 0;
let r = frontend::bind(
portal,
statement.name(),
param_formats,
params.zip(param_types).enumerate(),
|(idx, (param, ty)), buf| match param.to_sql_checked(ty, buf) {
Ok(IsNull::No) => Ok(postgres_protocol2::IsNull::No),
Ok(IsNull::Yes) => Ok(postgres_protocol2::IsNull::Yes),
Err(e) => {
error_idx = idx;
Err(e)
}
},
Some(1),
buf,
);
match r {
Ok(()) => Ok(()),
Err(frontend::BindError::Conversion(e)) => Err(Error::to_sql(e, error_idx)),
Err(frontend::BindError::Serialization(e)) => Err(Error::encode(e)),
}
}
pin_project! {
/// A stream of table rows.
pub struct RowStream {
statement: Statement,
responses: Responses,
command_tag: Option<String>,
output_format: Format,
status: ReadyForQueryStatus,
#[pin]
_p: PhantomPinned,
}
}
impl Stream for RowStream {
type Item = Result<Row, Error>;
fn poll_next(self: Pin<&mut Self>, cx: &mut Context<'_>) -> Poll<Option<Self::Item>> {
let this = self.get_mut();
let this = self.project();
loop {
match ready!(this.responses.poll_next(cx)?) {
Message::DataRow(body) => {
return Poll::Ready(Some(Ok(Row::new(
this.statement.clone(),
body,
this.output_format,
*this.output_format,
)?)));
}
Message::EmptyQueryResponse | Message::PortalSuspended => {}
Message::CommandComplete(body) => {
if let Ok(tag) = body.tag() {
this.command_tag = Some(tag.to_string());
*this.command_tag = Some(tag.to_string());
}
}
Message::ReadyForQuery(status) => {
this.status = status.into();
*this.status = status.into();
return Poll::Ready(None);
}
_ => return Poll::Ready(Some(Err(Error::unexpected_message()))),
@@ -148,3 +276,24 @@ impl Stream for RowStream<'_> {
}
}
}
impl RowStream {
/// Returns information about the columns of data in the row.
pub fn columns(&self) -> &[Column] {
self.statement.columns()
}
/// Returns the command tag of this query.
///
/// This is only available after the stream has been exhausted.
pub fn command_tag(&self) -> Option<String> {
self.command_tag.clone()
}
/// Returns if the connection is ready for querying, with the status of the connection.
///
/// This might be available only after the stream has been exhausted.
pub fn ready_status(&self) -> ReadyForQueryStatus {
self.status
}
}

View File

@@ -1,14 +1,19 @@
use std::marker::PhantomPinned;
use std::pin::Pin;
use std::sync::Arc;
use std::task::{Context, Poll};
use bytes::Bytes;
use fallible_iterator::FallibleIterator;
use futures_util::{Stream, ready};
use pin_project_lite::pin_project;
use postgres_protocol2::message::backend::Message;
use postgres_protocol2::message::frontend;
use tracing::debug;
use crate::client::{InnerClient, Responses};
use crate::codec::FrontendMessage;
use crate::connection::RequestMessages;
use crate::{Error, ReadyForQueryStatus, SimpleQueryMessage, SimpleQueryRow};
/// Information about a column of a single query row.
@@ -28,28 +33,28 @@ impl SimpleColumn {
}
}
pub async fn simple_query<'a>(
client: &'a mut InnerClient,
query: &str,
) -> Result<SimpleQueryStream<'a>, Error> {
pub async fn simple_query(client: &InnerClient, query: &str) -> Result<SimpleQueryStream, Error> {
debug!("executing simple query: {}", query);
let responses = client.send_simple_query(query)?;
let buf = encode(client, query)?;
let responses = client.send(RequestMessages::Single(FrontendMessage::Raw(buf)))?;
Ok(SimpleQueryStream {
responses,
columns: None,
status: ReadyForQueryStatus::Unknown,
_p: PhantomPinned,
})
}
pub async fn batch_execute(
client: &mut InnerClient,
client: &InnerClient,
query: &str,
) -> Result<ReadyForQueryStatus, Error> {
debug!("executing statement batch: {}", query);
let responses = client.send_simple_query(query)?;
let buf = encode(client, query)?;
let mut responses = client.send(RequestMessages::Single(FrontendMessage::Raw(buf)))?;
loop {
match responses.next().await? {
@@ -63,16 +68,25 @@ pub async fn batch_execute(
}
}
pub(crate) fn encode(client: &InnerClient, query: &str) -> Result<Bytes, Error> {
client.with_buf(|buf| {
frontend::query(query, buf).map_err(Error::encode)?;
Ok(buf.split().freeze())
})
}
pin_project! {
/// A stream of simple query results.
pub struct SimpleQueryStream<'a> {
responses: &'a mut Responses,
pub struct SimpleQueryStream {
responses: Responses,
columns: Option<Arc<[SimpleColumn]>>,
status: ReadyForQueryStatus,
#[pin]
_p: PhantomPinned,
}
}
impl SimpleQueryStream<'_> {
impl SimpleQueryStream {
/// Returns if the connection is ready for querying, with the status of the connection.
///
/// This might be available only after the stream has been exhausted.
@@ -81,7 +95,7 @@ impl SimpleQueryStream<'_> {
}
}
impl Stream for SimpleQueryStream<'_> {
impl Stream for SimpleQueryStream {
type Item = Result<SimpleQueryMessage, Error>;
fn poll_next(self: Pin<&mut Self>, cx: &mut Context<'_>) -> Poll<Option<Self::Item>> {

View File

@@ -1,15 +1,35 @@
use std::fmt;
use std::sync::Arc;
use std::sync::{Arc, Weak};
use crate::types::Type;
use postgres_protocol2::Oid;
use postgres_protocol2::message::backend::Field;
use postgres_protocol2::message::frontend;
use crate::client::InnerClient;
use crate::codec::FrontendMessage;
use crate::connection::RequestMessages;
use crate::types::Type;
struct StatementInner {
client: Weak<InnerClient>,
name: &'static str,
params: Vec<Type>,
columns: Vec<Column>,
}
impl Drop for StatementInner {
fn drop(&mut self) {
if let Some(client) = self.client.upgrade() {
let buf = client.with_buf(|buf| {
frontend::close(b'S', self.name, buf).unwrap();
frontend::sync(buf);
buf.split().freeze()
});
let _ = client.send(RequestMessages::Single(FrontendMessage::Raw(buf)));
}
}
}
/// A prepared statement.
///
/// Prepared statements can only be used with the connection that created them.
@@ -17,14 +37,38 @@ struct StatementInner {
pub struct Statement(Arc<StatementInner>);
impl Statement {
pub(crate) fn new(name: &'static str, columns: Vec<Column>) -> Statement {
Statement(Arc::new(StatementInner { name, columns }))
pub(crate) fn new(
inner: &Arc<InnerClient>,
name: &'static str,
params: Vec<Type>,
columns: Vec<Column>,
) -> Statement {
Statement(Arc::new(StatementInner {
client: Arc::downgrade(inner),
name,
params,
columns,
}))
}
pub(crate) fn new_anonymous(params: Vec<Type>, columns: Vec<Column>) -> Statement {
Statement(Arc::new(StatementInner {
client: Weak::new(),
name: "<anonymous>",
params,
columns,
}))
}
pub(crate) fn name(&self) -> &str {
self.0.name
}
/// Returns the expected types of the statement's parameters.
pub fn params(&self) -> &[Type] {
&self.0.params
}
/// Returns information about the columns returned when the statement is queried.
pub fn columns(&self) -> &[Column] {
&self.0.columns
@@ -34,7 +78,7 @@ impl Statement {
/// Information about a column of a query.
pub struct Column {
name: String,
pub(crate) type_: Type,
type_: Type,
// raw fields from RowDescription
table_oid: Oid,

View File

@@ -1,3 +1,7 @@
use postgres_protocol2::message::frontend;
use crate::codec::FrontendMessage;
use crate::connection::RequestMessages;
use crate::query::RowStream;
use crate::{CancelToken, Client, Error, ReadyForQueryStatus};
@@ -16,7 +20,14 @@ impl Drop for Transaction<'_> {
return;
}
let _ = self.client.inner_mut().send_simple_query("ROLLBACK");
let buf = self.client.inner().with_buf(|buf| {
frontend::query("ROLLBACK", buf).unwrap();
buf.split().freeze()
});
let _ = self
.client
.inner()
.send(RequestMessages::Single(FrontendMessage::Raw(buf)));
}
}
@@ -43,11 +54,7 @@ impl<'a> Transaction<'a> {
}
/// Like `Client::query_raw_txt`.
pub async fn query_raw_txt<S, I>(
&mut self,
statement: &str,
params: I,
) -> Result<RowStream, Error>
pub async fn query_raw_txt<S, I>(&self, statement: &str, params: I) -> Result<RowStream, Error>
where
S: AsRef<str>,
I: IntoIterator<Item = Option<S>>,

View File

@@ -1,7 +1,7 @@
use std::borrow::Cow;
use std::fs::{self, File};
use std::io::{self, Write};
use std::os::fd::AsFd;
use std::os::fd::AsRawFd;
use camino::{Utf8Path, Utf8PathBuf};
@@ -210,13 +210,13 @@ pub fn overwrite(
/// Syncs the filesystem for the given file descriptor.
#[cfg_attr(target_os = "macos", allow(unused_variables))]
pub fn syncfs(fd: impl AsFd) -> anyhow::Result<()> {
pub fn syncfs(fd: impl AsRawFd) -> anyhow::Result<()> {
// Linux guarantees durability for syncfs.
// POSIX doesn't have syncfs, and further does not actually guarantee durability of sync().
#[cfg(target_os = "linux")]
{
use anyhow::Context;
nix::unistd::syncfs(fd).context("syncfs")?;
nix::unistd::syncfs(fd.as_raw_fd()).context("syncfs")?;
}
#[cfg(target_os = "macos")]
{

View File

@@ -11,9 +11,9 @@ pub fn rename_noreplace<P1: ?Sized + NixPath, P2: ?Sized + NixPath>(
#[cfg(all(target_os = "linux", target_env = "gnu"))]
{
nix::fcntl::renameat2(
nix::fcntl::AT_FDCWD,
None,
src,
nix::fcntl::AT_FDCWD,
None,
dst,
nix::fcntl::RenameFlags::RENAME_NOREPLACE,
)

View File

@@ -1,6 +1,6 @@
//! A module to create and read lock files.
//!
//! File locking is done using [`nix::fcntl::Flock`] exclusive locks.
//! File locking is done using [`fcntl::flock`] exclusive locks.
//! The only consumer of this module is currently
//! [`pid_file`](crate::pid_file). See the module-level comment
//! there for potential pitfalls with lock files that are used
@@ -9,25 +9,26 @@
use std::fs;
use std::io::{Read, Write};
use std::ops::Deref;
use std::os::unix::prelude::AsRawFd;
use anyhow::Context;
use camino::{Utf8Path, Utf8PathBuf};
use nix::errno::Errno::EAGAIN;
use nix::fcntl::{Flock, FlockArg};
use nix::fcntl;
use crate::crashsafe;
/// A handle to an open and flocked, but not-yet-written lock file.
/// A handle to an open and unlocked, but not-yet-written lock file.
/// Returned by [`create_exclusive`].
#[must_use]
pub struct UnwrittenLockFile {
path: Utf8PathBuf,
file: Flock<fs::File>,
file: fs::File,
}
/// Returned by [`UnwrittenLockFile::write_content`].
#[must_use]
pub struct LockFileGuard(Flock<fs::File>);
pub struct LockFileGuard(fs::File);
impl Deref for LockFileGuard {
type Target = fs::File;
@@ -66,14 +67,17 @@ pub fn create_exclusive(lock_file_path: &Utf8Path) -> anyhow::Result<UnwrittenLo
.open(lock_file_path)
.context("open lock file")?;
let res = Flock::lock(lock_file, FlockArg::LockExclusiveNonblock);
let res = fcntl::flock(
lock_file.as_raw_fd(),
fcntl::FlockArg::LockExclusiveNonblock,
);
match res {
Ok(lock_file) => Ok(UnwrittenLockFile {
Ok(()) => Ok(UnwrittenLockFile {
path: lock_file_path.to_owned(),
file: lock_file,
}),
Err((_, EAGAIN)) => anyhow::bail!("file is already locked"),
Err((_, e)) => Err(e).context("flock error"),
Err(EAGAIN) => anyhow::bail!("file is already locked"),
Err(e) => Err(e).context("flock error"),
}
}
@@ -101,37 +105,32 @@ pub enum LockFileRead {
/// Check the [`LockFileRead`] variants for details.
pub fn read_and_hold_lock_file(path: &Utf8Path) -> anyhow::Result<LockFileRead> {
let res = fs::OpenOptions::new().read(true).open(path);
let lock_file = match res {
let mut lock_file = match res {
Ok(f) => f,
Err(e) => match e.kind() {
std::io::ErrorKind::NotFound => return Ok(LockFileRead::NotExist),
_ => return Err(e).context("open lock file"),
},
};
let res = Flock::lock(lock_file, FlockArg::LockExclusiveNonblock);
let res = fcntl::flock(
lock_file.as_raw_fd(),
fcntl::FlockArg::LockExclusiveNonblock,
);
// We need the content regardless of lock success / failure.
// But, read it after flock so that, if it succeeded, the content is consistent.
let mut content = String::new();
lock_file
.read_to_string(&mut content)
.context("read lock file")?;
match res {
Ok(mut locked_file) => {
let mut content = String::new();
locked_file
.read_to_string(&mut content)
.context("read lock file")?;
Ok(LockFileRead::NotHeldByAnyProcess(
LockFileGuard(locked_file),
content,
))
}
Err((mut not_locked_file, EAGAIN)) => {
let mut content = String::new();
not_locked_file
.read_to_string(&mut content)
.context("read lock file")?;
Ok(LockFileRead::LockedByOtherProcess {
not_locked_file,
content,
})
}
Err((_, e)) => Err(e).context("flock error"),
Ok(()) => Ok(LockFileRead::NotHeldByAnyProcess(
LockFileGuard(lock_file),
content,
)),
Err(EAGAIN) => Ok(LockFileRead::LockedByOtherProcess {
not_locked_file: lock_file,
content,
}),
Err(e) => Err(e).context("flock error"),
}
}

View File

@@ -127,12 +127,12 @@ macro_rules! __check_fields_present {
match check_fields_present0($extractors) {
Ok(FoundEverything) => Ok(()),
Ok(Unconfigured) if cfg!(feature = "testing") => {
Ok(Unconfigured) if cfg!(test) => {
// allow unconfigured in tests
Ok(())
},
Ok(Unconfigured) => {
panic!(r#"utils::tracing_span_assert: outside of #[cfg(feature = "testing")] expected tracing to be configured with tracing_error::ErrorLayer"#)
panic!("utils::tracing_span_assert: outside of #[cfg(test)] expected tracing to be configured with tracing_error::ErrorLayer")
},
Err(missing) => Err(missing)
}

View File

@@ -1,7 +1,6 @@
#![allow(clippy::todo)]
use std::ffi::CString;
use std::str::FromStr;
use postgres_ffi::WAL_SEGMENT_SIZE;
use utils::id::TenantTimelineId;
@@ -174,8 +173,6 @@ pub struct Config {
pub ttid: TenantTimelineId,
/// List of safekeepers in format `host:port`
pub safekeepers_list: Vec<String>,
/// libpq connection info options
pub safekeeper_conninfo_options: String,
/// Safekeeper reconnect timeout in milliseconds
pub safekeeper_reconnect_timeout: i32,
/// Safekeeper connection timeout in milliseconds
@@ -205,9 +202,6 @@ impl Wrapper {
.into_bytes_with_nul();
assert!(safekeepers_list_vec.len() == safekeepers_list_vec.capacity());
let safekeepers_list = safekeepers_list_vec.as_mut_ptr() as *mut std::ffi::c_char;
let safekeeper_conninfo_options = CString::from_str(&config.safekeeper_conninfo_options)
.unwrap()
.into_raw();
let callback_data = Box::into_raw(Box::new(api)) as *mut ::std::os::raw::c_void;
@@ -215,7 +209,6 @@ impl Wrapper {
neon_tenant,
neon_timeline,
safekeepers_list,
safekeeper_conninfo_options,
safekeeper_reconnect_timeout: config.safekeeper_reconnect_timeout,
safekeeper_connection_timeout: config.safekeeper_connection_timeout,
wal_segment_size: WAL_SEGMENT_SIZE as i32, // default 16MB
@@ -583,7 +576,6 @@ mod tests {
let config = crate::walproposer::Config {
ttid,
safekeepers_list: vec!["localhost:5000".to_string()],
safekeeper_conninfo_options: String::new(),
safekeeper_reconnect_timeout: 1000,
safekeeper_connection_timeout: 10000,
sync_safekeepers: true,

View File

@@ -17,69 +17,49 @@ anyhow.workspace = true
arc-swap.workspace = true
async-compression.workspace = true
async-stream.workspace = true
bincode.workspace = true
bit_field.workspace = true
bincode.workspace = true
byteorder.workspace = true
bytes.workspace = true
camino-tempfile.workspace = true
camino.workspace = true
camino-tempfile.workspace = true
chrono = { workspace = true, features = ["serde"] }
clap = { workspace = true, features = ["string"] }
consumption_metrics.workspace = true
crc32c.workspace = true
either.workspace = true
enum-map.workspace = true
enumset = { workspace = true, features = ["serde"]}
fail.workspace = true
futures.workspace = true
hashlink.workspace = true
hex.workspace = true
http-utils.workspace = true
humantime-serde.workspace = true
humantime.workspace = true
humantime-serde.workspace = true
hyper0.workspace = true
itertools.workspace = true
jsonwebtoken.workspace = true
md5.workspace = true
metrics.workspace = true
nix.workspace = true
num_cpus.workspace = true # hack to get the number of worker threads tokio uses
# hack to get the number of worker threads tokio uses
num_cpus.workspace = true
num-traits.workspace = true
once_cell.workspace = true
pageserver_api.workspace = true
pageserver_client.workspace = true # for ResponseErrorMessageExt TOOD refactor that
pageserver_compaction.workspace = true
pageserver_page_api.workspace = true
pem.workspace = true
pin-project-lite.workspace = true
postgres_backend.workspace = true
postgres_connection.workspace = true
postgres_ffi.workspace = true
postgres_initdb.workspace = true
postgres-protocol.workspace = true
postgres-types.workspace = true
posthog_client_lite.workspace = true
postgres_initdb.workspace = true
pprof.workspace = true
pq_proto.workspace = true
rand.workspace = true
range-set-blaze = { version = "0.1.16", features = ["alloc"] }
regex.workspace = true
remote_storage.workspace = true
reqwest.workspace = true
rpds.workspace = true
rustls.workspace = true
scopeguard.workspace = true
send-future.workspace = true
serde.workspace = true
serde_json = { workspace = true, features = ["raw_value"] }
serde_path_to_error.workspace = true
serde_with.workspace = true
serde.workspace = true
smallvec.workspace = true
storage_broker.workspace = true
strum_macros.workspace = true
strum.workspace = true
sysinfo.workspace = true
tenant_size_model.workspace = true
tokio-tar.workspace = true
thiserror.workspace = true
tikv-jemallocator.workspace = true
tokio = { workspace = true, features = ["process", "sync", "fs", "rt", "io-util", "time"] }
@@ -88,19 +68,34 @@ tokio-io-timeout.workspace = true
tokio-postgres.workspace = true
tokio-rustls.workspace = true
tokio-stream.workspace = true
tokio-tar.workspace = true
tokio-util.workspace = true
toml_edit = { workspace = true, features = [ "serde" ] }
tonic.workspace = true
tonic-reflection.workspace = true
tracing.workspace = true
tracing-utils.workspace = true
url.workspace = true
utils.workspace = true
wal_decoder.workspace = true
walkdir.workspace = true
metrics.workspace = true
pageserver_api.workspace = true
pageserver_client.workspace = true # for ResponseErrorMessageExt TOOD refactor that
pageserver_compaction.workspace = true
pem.workspace = true
postgres_connection.workspace = true
postgres_ffi.workspace = true
pq_proto.workspace = true
remote_storage.workspace = true
storage_broker.workspace = true
tenant_size_model.workspace = true
http-utils.workspace = true
utils.workspace = true
workspace_hack.workspace = true
twox-hash.workspace = true
reqwest.workspace = true
rpds.workspace = true
enum-map.workspace = true
enumset = { workspace = true, features = ["serde"]}
strum.workspace = true
strum_macros.workspace = true
wal_decoder.workspace = true
smallvec.workspace = true
[target.'cfg(target_os = "linux")'.dependencies]
procfs.workspace = true

View File

@@ -264,56 +264,10 @@ mod propagation_of_cached_label_value {
}
}
criterion_group!(histograms, histograms::bench_bucket_scalability);
mod histograms {
use std::time::Instant;
use criterion::{BenchmarkId, Criterion};
use metrics::core::Collector;
pub fn bench_bucket_scalability(c: &mut Criterion) {
let mut g = c.benchmark_group("bucket_scalability");
for n in [1, 4, 8, 16, 32, 64, 128, 256] {
g.bench_with_input(BenchmarkId::new("nbuckets", n), &n, |b, n| {
b.iter_custom(|iters| {
let buckets: Vec<f64> = (0..*n).map(|i| i as f64 * 100.0).collect();
let histo = metrics::Histogram::with_opts(
metrics::prometheus::HistogramOpts::new("name", "help")
.buckets(buckets.clone()),
)
.unwrap();
let start = Instant::now();
for i in 0..usize::try_from(iters).unwrap() {
histo.observe(buckets[i % buckets.len()]);
}
let elapsed = start.elapsed();
// self-test
let mfs = histo.collect();
assert_eq!(mfs.len(), 1);
let metrics = mfs[0].get_metric();
assert_eq!(metrics.len(), 1);
let histo = metrics[0].get_histogram();
let buckets = histo.get_bucket();
assert!(
buckets
.iter()
.enumerate()
.all(|(i, b)| b.get_cumulative_count()
>= i as u64 * (iters / buckets.len() as u64))
);
elapsed
})
});
}
}
}
criterion_main!(
label_values,
single_metric_multicore_scalability,
propagation_of_cached_label_value,
histograms,
propagation_of_cached_label_value
);
/*
@@ -336,14 +290,6 @@ propagation_of_cached_label_value__naive/nthreads/8 time: [211.50 ns 214.44 ns
propagation_of_cached_label_value__long_lived_reference_per_thread/nthreads/1 time: [14.135 ns 14.147 ns 14.160 ns]
propagation_of_cached_label_value__long_lived_reference_per_thread/nthreads/4 time: [14.243 ns 14.255 ns 14.268 ns]
propagation_of_cached_label_value__long_lived_reference_per_thread/nthreads/8 time: [14.470 ns 14.682 ns 14.895 ns]
bucket_scalability/nbuckets/1 time: [30.352 ns 30.353 ns 30.354 ns]
bucket_scalability/nbuckets/4 time: [30.464 ns 30.465 ns 30.467 ns]
bucket_scalability/nbuckets/8 time: [30.569 ns 30.575 ns 30.584 ns]
bucket_scalability/nbuckets/16 time: [30.961 ns 30.965 ns 30.969 ns]
bucket_scalability/nbuckets/32 time: [35.691 ns 35.707 ns 35.722 ns]
bucket_scalability/nbuckets/64 time: [47.829 ns 47.898 ns 47.974 ns]
bucket_scalability/nbuckets/128 time: [73.479 ns 73.512 ns 73.545 ns]
bucket_scalability/nbuckets/256 time: [127.92 ns 127.94 ns 127.96 ns]
Results on an i3en.3xlarge instance
@@ -398,14 +344,6 @@ propagation_of_cached_label_value__naive/nthreads/8 time: [434.87 ns 456.4
propagation_of_cached_label_value__long_lived_reference_per_thread/nthreads/1 time: [3.3767 ns 3.3974 ns 3.4220 ns]
propagation_of_cached_label_value__long_lived_reference_per_thread/nthreads/4 time: [3.6105 ns 4.2355 ns 5.1463 ns]
propagation_of_cached_label_value__long_lived_reference_per_thread/nthreads/8 time: [4.0889 ns 4.9714 ns 6.0779 ns]
bucket_scalability/nbuckets/1 time: [4.8455 ns 4.8542 ns 4.8646 ns]
bucket_scalability/nbuckets/4 time: [4.5663 ns 4.5722 ns 4.5787 ns]
bucket_scalability/nbuckets/8 time: [4.5531 ns 4.5670 ns 4.5842 ns]
bucket_scalability/nbuckets/16 time: [4.6392 ns 4.6524 ns 4.6685 ns]
bucket_scalability/nbuckets/32 time: [6.0302 ns 6.0439 ns 6.0589 ns]
bucket_scalability/nbuckets/64 time: [10.608 ns 10.644 ns 10.691 ns]
bucket_scalability/nbuckets/128 time: [22.178 ns 22.316 ns 22.483 ns]
bucket_scalability/nbuckets/256 time: [42.190 ns 42.328 ns 42.492 ns]
Results on a Hetzner AX102 AMD Ryzen 9 7950X3D 16-Core Processor
@@ -424,13 +362,5 @@ propagation_of_cached_label_value__naive/nthreads/8 time: [164.24 ns 170.1
propagation_of_cached_label_value__long_lived_reference_per_thread/nthreads/1 time: [2.2915 ns 2.2960 ns 2.3012 ns]
propagation_of_cached_label_value__long_lived_reference_per_thread/nthreads/4 time: [2.5726 ns 2.6158 ns 2.6624 ns]
propagation_of_cached_label_value__long_lived_reference_per_thread/nthreads/8 time: [2.7068 ns 2.8243 ns 2.9824 ns]
bucket_scalability/nbuckets/1 time: [6.3998 ns 6.4288 ns 6.4684 ns]
bucket_scalability/nbuckets/4 time: [6.3603 ns 6.3620 ns 6.3637 ns]
bucket_scalability/nbuckets/8 time: [6.1646 ns 6.1654 ns 6.1667 ns]
bucket_scalability/nbuckets/16 time: [6.1341 ns 6.1391 ns 6.1454 ns]
bucket_scalability/nbuckets/32 time: [8.2206 ns 8.2254 ns 8.2301 ns]
bucket_scalability/nbuckets/64 time: [13.988 ns 13.994 ns 14.000 ns]
bucket_scalability/nbuckets/128 time: [28.180 ns 28.216 ns 28.251 ns]
bucket_scalability/nbuckets/256 time: [54.914 ns 54.931 ns 54.951 ns]
*/

View File

@@ -1,19 +0,0 @@
[package]
name = "pageserver_page_api"
version = "0.1.0"
edition.workspace = true
license.workspace = true
[dependencies]
bytes.workspace = true
pageserver_api.workspace = true
postgres_ffi.workspace = true
prost.workspace = true
smallvec.workspace = true
thiserror.workspace = true
tonic.workspace = true
utils.workspace = true
workspace_hack.workspace = true
[build-dependencies]
tonic-build.workspace = true

View File

@@ -1,13 +0,0 @@
use std::env;
use std::path::PathBuf;
/// Generates Rust code from .proto Protobuf schemas, along with a binary file
/// descriptor set for Protobuf schema reflection.
fn main() -> Result<(), Box<dyn std::error::Error>> {
let out_dir = PathBuf::from(env::var("OUT_DIR")?);
tonic_build::configure()
.bytes(["."])
.file_descriptor_set_path(out_dir.join("page_api_descriptor.bin"))
.compile_protos(&["proto/page_service.proto"], &["proto"])
.map_err(|err| err.into())
}

View File

@@ -1,239 +0,0 @@
// Page service, presented by pageservers for computes.
//
// This is the compute read path. It primarily serves page versions at given
// LSNs, but also base backups, SLRU segments, and relation metadata.
//
// EXPERIMENTAL: this is still under development and subject to change.
//
// Request metadata headers:
// - authorization: JWT token ("Bearer <token>"), if auth is enabled
// - neon-tenant-id: tenant ID ("7c4a1f9e3bd6470c8f3e21a65bd2e980")
// - neon-shard-id: shard ID, as <number><count> in hex ("0b10" = shard 11 of 16, 0-based)
// - neon-timeline-id: timeline ID ("f08c4e9a2d5f76b1e3a7c2d8910f4b3e")
//
// The service can be accessed via e.g. grpcurl:
//
// ```
// grpcurl \
// -plaintext \
// -H "neon-tenant-id: 7c4a1f9e3bd6470c8f3e21a65bd2e980" \
// -H "neon-shard-id: 0b10" \
// -H "neon-timeline-id: f08c4e9a2d5f76b1e3a7c2d8910f4b3e" \
// -H "authorization: Bearer $JWT" \
// -d '{"read_lsn": {"request_lsn": 1234567890}, "rel": {"spc_oid": 1663, "db_oid": 1234, "rel_number": 5678, "fork_number": 0}}'
// localhost:51051 page_api.PageService/CheckRelExists
// ```
//
// TODO: consider adding neon-compute-mode ("primary", "static", "replica").
// However, this will require reconnecting when changing modes.
//
// TODO: write implementation guidance on
// - Health checks
// - Tracing, OpenTelemetry
// - Compression
syntax = "proto3";
package page_api;
service PageService {
// Returns whether a relation exists.
rpc CheckRelExists(CheckRelExistsRequest) returns (CheckRelExistsResponse);
// Fetches a base backup.
rpc GetBaseBackup (GetBaseBackupRequest) returns (stream GetBaseBackupResponseChunk);
// Returns the total size of a database, as # of bytes.
rpc GetDbSize (GetDbSizeRequest) returns (GetDbSizeResponse);
// Fetches pages.
//
// This is implemented as a bidirectional streaming RPC for performance. Unary
// requests incur costs for e.g. HTTP/2 stream setup, header parsing,
// authentication, and so on -- with streaming, we only pay these costs during
// the initial stream setup. This ~doubles throughput in benchmarks. Other
// RPCs use regular unary requests, since they are not as frequent and
// performance-critical, and this simplifies implementation.
//
// NB: a gRPC status response (e.g. errors) will terminate the stream. The
// stream may be shared by multiple Postgres backends, so we avoid this by
// sending them as GetPageResponse.status_code instead.
rpc GetPages (stream GetPageRequest) returns (stream GetPageResponse);
// Returns the size of a relation, as # of blocks.
rpc GetRelSize (GetRelSizeRequest) returns (GetRelSizeResponse);
// Fetches an SLRU segment.
rpc GetSlruSegment (GetSlruSegmentRequest) returns (GetSlruSegmentResponse);
}
// The LSN a request should read at.
message ReadLsn {
// The request's read LSN. Required.
uint64 request_lsn = 1;
// If given, the caller guarantees that the page has not been modified since
// this LSN. Must be smaller than or equal to request_lsn. This allows the
// Pageserver to serve an old page without waiting for the request LSN to
// arrive. Valid for all request types.
//
// It is undefined behaviour to make a request such that the page was, in
// fact, modified between request_lsn and not_modified_since_lsn. The
// Pageserver might detect it and return an error, or it might return the old
// page version or the new page version. Setting not_modified_since_lsn equal
// to request_lsn is always safe, but can lead to unnecessary waiting.
uint64 not_modified_since_lsn = 2;
}
// A relation identifier.
message RelTag {
uint32 spc_oid = 1;
uint32 db_oid = 2;
uint32 rel_number = 3;
uint32 fork_number = 4;
}
// Checks whether a relation exists, at the given LSN. Only valid on shard 0,
// other shards will error.
message CheckRelExistsRequest {
ReadLsn read_lsn = 1;
RelTag rel = 2;
}
message CheckRelExistsResponse {
bool exists = 1;
}
// Requests a base backup at a given LSN.
message GetBaseBackupRequest {
// The LSN to fetch a base backup at.
ReadLsn read_lsn = 1;
// If true, logical replication slots will not be created.
bool replica = 2;
}
// Base backup response chunk, returned as an ordered stream.
message GetBaseBackupResponseChunk {
// A basebackup data chunk. The size is undefined, but bounded by the 4 MB
// gRPC message size limit.
bytes chunk = 1;
}
// Requests the size of a database, as # of bytes. Only valid on shard 0, other
// shards will error.
message GetDbSizeRequest {
ReadLsn read_lsn = 1;
uint32 db_oid = 2;
}
message GetDbSizeResponse {
uint64 num_bytes = 1;
}
// Requests one or more pages.
message GetPageRequest {
// A request ID. Will be included in the response. Should be unique for
// in-flight requests on the stream.
uint64 request_id = 1;
// The request class.
GetPageClass request_class = 2;
// The LSN to read at.
ReadLsn read_lsn = 3;
// The relation to read from.
RelTag rel = 4;
// Page numbers to read. Must belong to the remote shard.
//
// Multiple pages will be executed as a single batch by the Pageserver,
// amortizing layer access costs and parallelizing them. This may increase the
// latency of any individual request, but improves the overall latency and
// throughput of the batch as a whole.
//
// TODO: this causes an allocation in the common single-block case. The sender
// can use a SmallVec to stack-allocate it, but Prost will always deserialize
// into a heap-allocated Vec. Consider optimizing this.
//
// TODO: we might be able to avoid a sort or something if we mandate that these
// are always in order. But we can't currenly rely on this on the server, because
// of compatibility with the libpq protocol handler.
repeated uint32 block_number = 5;
}
// A GetPageRequest class. Primarily intended for observability, but may also be
// used for prioritization in the future.
enum GetPageClass {
// Unknown class. For backwards compatibility: used when an older client version sends a class
// that a newer server version has removed.
GET_PAGE_CLASS_UNKNOWN = 0;
// A normal request. This is the default.
GET_PAGE_CLASS_NORMAL = 1;
// A prefetch request. NB: can only be classified on pg < 18.
GET_PAGE_CLASS_PREFETCH = 2;
// A background request (e.g. vacuum).
GET_PAGE_CLASS_BACKGROUND = 3;
}
// A GetPage response.
//
// A batch response will contain all of the requested pages. We could eagerly
// emit individual pages as soon as they are ready, but on a readv() Postgres
// holds buffer pool locks on all pages in the batch and we'll only return once
// the entire batch is ready, so no one can make use of the individual pages.
message GetPageResponse {
// The original request's ID.
uint64 request_id = 1;
// The response status code.
GetPageStatusCode status_code = 2;
// A string describing the status, if any.
string reason = 3;
// The 8KB page images, in the same order as the request. Empty if status_code != OK.
repeated bytes page_image = 4;
}
// A GetPageResponse status code.
//
// These are effectively equivalent to gRPC statuses. However, we use a bidirectional stream
// (potentially shared by many backends), and a gRPC status response would terminate the stream so
// we send GetPageResponse messages with these codes instead.
enum GetPageStatusCode {
// Unknown status. For forwards compatibility: used when an older client version receives a new
// status code from a newer server version.
GET_PAGE_STATUS_CODE_UNKNOWN = 0;
// The request was successful.
GET_PAGE_STATUS_CODE_OK = 1;
// The page did not exist. The tenant/timeline/shard has already been
// validated during stream setup.
GET_PAGE_STATUS_CODE_NOT_FOUND = 2;
// The request was invalid.
GET_PAGE_STATUS_CODE_INVALID_REQUEST = 3;
// The request failed due to an internal server error.
GET_PAGE_STATUS_CODE_INTERNAL_ERROR = 4;
// The tenant is rate limited. Slow down and retry later.
GET_PAGE_STATUS_CODE_SLOW_DOWN = 5;
// NB: shutdown errors are emitted as a gRPC Unavailable status.
//
// TODO: consider adding a GET_PAGE_STATUS_CODE_LAYER_DOWNLOAD in the case of a layer download.
// This could free up the server task to process other requests while the download is in progress.
}
// Fetches the size of a relation at a given LSN, as # of blocks. Only valid on
// shard 0, other shards will error.
message GetRelSizeRequest {
ReadLsn read_lsn = 1;
RelTag rel = 2;
}
message GetRelSizeResponse {
uint32 num_blocks = 1;
}
// Requests an SLRU segment. Only valid on shard 0, other shards will error.
message GetSlruSegmentRequest {
ReadLsn read_lsn = 1;
uint32 kind = 2;
uint32 segno = 3;
}
// Returns an SLRU segment.
//
// These are up 32 pages (256 KB), so we can send them as a single response.
message GetSlruSegmentResponse {
bytes segment = 1;
}

View File

@@ -1,23 +0,0 @@
//! This crate provides the Pageserver's page API. It contains:
//!
//! * proto/page_service.proto: the Protobuf schema for the page API.
//! * proto: auto-generated Protobuf types for gRPC.
//!
//! This crate is used by both the client and the server. Try to keep it slim.
// Code generated by protobuf.
pub mod proto {
tonic::include_proto!("page_api");
/// File descriptor set for Protobuf schema reflection. This allows using
/// e.g. grpcurl with the API.
pub const FILE_DESCRIPTOR_SET: &[u8] =
tonic::include_file_descriptor_set!("page_api_descriptor");
pub use page_service_client::PageServiceClient;
pub use page_service_server::{PageService, PageServiceServer};
}
mod model;
pub use model::*;

View File

@@ -1,595 +0,0 @@
//! Structs representing the canonical page service API.
//!
//! These mirror the autogenerated Protobuf types. The differences are:
//!
//! - Types that are in fact required by the API are not Options. The protobuf "required"
//! attribute is deprecated and 'prost' marks a lot of members as optional because of that.
//! (See <https://github.com/tokio-rs/prost/issues/800> for a gripe on this)
//!
//! - Use more precise datatypes, e.g. Lsn and uints shorter than 32 bits.
//!
//! - Validate protocol invariants, via try_from() and try_into().
use bytes::Bytes;
use postgres_ffi::Oid;
use smallvec::SmallVec;
// TODO: split out Lsn, RelTag, SlruKind, Oid and other basic types to a separate crate, to avoid
// pulling in all of their other crate dependencies when building the client.
use utils::lsn::Lsn;
use crate::proto;
/// A protocol error. Typically returned via try_from() or try_into().
#[derive(thiserror::Error, Debug)]
pub enum ProtocolError {
#[error("field '{0}' has invalid value '{1}'")]
Invalid(&'static str, String),
#[error("required field '{0}' is missing")]
Missing(&'static str),
}
impl ProtocolError {
/// Helper to generate a new ProtocolError::Invalid for the given field and value.
pub fn invalid(field: &'static str, value: impl std::fmt::Debug) -> Self {
Self::Invalid(field, format!("{value:?}"))
}
}
impl From<ProtocolError> for tonic::Status {
fn from(err: ProtocolError) -> Self {
tonic::Status::invalid_argument(format!("{err}"))
}
}
/// The LSN a request should read at.
#[derive(Clone, Copy, Debug)]
pub struct ReadLsn {
/// The request's read LSN.
pub request_lsn: Lsn,
/// If given, the caller guarantees that the page has not been modified since this LSN. Must be
/// smaller than or equal to request_lsn. This allows the Pageserver to serve an old page
/// without waiting for the request LSN to arrive. Valid for all request types.
///
/// It is undefined behaviour to make a request such that the page was, in fact, modified
/// between request_lsn and not_modified_since_lsn. The Pageserver might detect it and return an
/// error, or it might return the old page version or the new page version. Setting
/// not_modified_since_lsn equal to request_lsn is always safe, but can lead to unnecessary
/// waiting.
pub not_modified_since_lsn: Option<Lsn>,
}
impl ReadLsn {
/// Validates the ReadLsn.
pub fn validate(&self) -> Result<(), ProtocolError> {
if self.request_lsn == Lsn::INVALID {
return Err(ProtocolError::invalid("request_lsn", self.request_lsn));
}
if self.not_modified_since_lsn > Some(self.request_lsn) {
return Err(ProtocolError::invalid(
"not_modified_since_lsn",
self.not_modified_since_lsn,
));
}
Ok(())
}
}
impl TryFrom<proto::ReadLsn> for ReadLsn {
type Error = ProtocolError;
fn try_from(pb: proto::ReadLsn) -> Result<Self, Self::Error> {
let read_lsn = Self {
request_lsn: Lsn(pb.request_lsn),
not_modified_since_lsn: match pb.not_modified_since_lsn {
0 => None,
lsn => Some(Lsn(lsn)),
},
};
read_lsn.validate()?;
Ok(read_lsn)
}
}
impl TryFrom<ReadLsn> for proto::ReadLsn {
type Error = ProtocolError;
fn try_from(read_lsn: ReadLsn) -> Result<Self, Self::Error> {
read_lsn.validate()?;
Ok(Self {
request_lsn: read_lsn.request_lsn.0,
not_modified_since_lsn: read_lsn.not_modified_since_lsn.unwrap_or_default().0,
})
}
}
// RelTag is defined in pageserver_api::reltag.
pub type RelTag = pageserver_api::reltag::RelTag;
impl TryFrom<proto::RelTag> for RelTag {
type Error = ProtocolError;
fn try_from(pb: proto::RelTag) -> Result<Self, Self::Error> {
Ok(Self {
spcnode: pb.spc_oid,
dbnode: pb.db_oid,
relnode: pb.rel_number,
forknum: pb
.fork_number
.try_into()
.map_err(|_| ProtocolError::invalid("fork_number", pb.fork_number))?,
})
}
}
impl From<RelTag> for proto::RelTag {
fn from(rel_tag: RelTag) -> Self {
Self {
spc_oid: rel_tag.spcnode,
db_oid: rel_tag.dbnode,
rel_number: rel_tag.relnode,
fork_number: rel_tag.forknum as u32,
}
}
}
/// Checks whether a relation exists, at the given LSN. Only valid on shard 0, other shards error.
#[derive(Clone, Copy, Debug)]
pub struct CheckRelExistsRequest {
pub read_lsn: ReadLsn,
pub rel: RelTag,
}
impl TryFrom<proto::CheckRelExistsRequest> for CheckRelExistsRequest {
type Error = ProtocolError;
fn try_from(pb: proto::CheckRelExistsRequest) -> Result<Self, Self::Error> {
Ok(Self {
read_lsn: pb
.read_lsn
.ok_or(ProtocolError::Missing("read_lsn"))?
.try_into()?,
rel: pb.rel.ok_or(ProtocolError::Missing("rel"))?.try_into()?,
})
}
}
pub type CheckRelExistsResponse = bool;
impl From<proto::CheckRelExistsResponse> for CheckRelExistsResponse {
fn from(pb: proto::CheckRelExistsResponse) -> Self {
pb.exists
}
}
impl From<CheckRelExistsResponse> for proto::CheckRelExistsResponse {
fn from(exists: CheckRelExistsResponse) -> Self {
Self { exists }
}
}
/// Requests a base backup at a given LSN.
#[derive(Clone, Copy, Debug)]
pub struct GetBaseBackupRequest {
/// The LSN to fetch a base backup at.
pub read_lsn: ReadLsn,
/// If true, logical replication slots will not be created.
pub replica: bool,
}
impl TryFrom<proto::GetBaseBackupRequest> for GetBaseBackupRequest {
type Error = ProtocolError;
fn try_from(pb: proto::GetBaseBackupRequest) -> Result<Self, Self::Error> {
Ok(Self {
read_lsn: pb
.read_lsn
.ok_or(ProtocolError::Missing("read_lsn"))?
.try_into()?,
replica: pb.replica,
})
}
}
impl TryFrom<GetBaseBackupRequest> for proto::GetBaseBackupRequest {
type Error = ProtocolError;
fn try_from(request: GetBaseBackupRequest) -> Result<Self, Self::Error> {
Ok(Self {
read_lsn: Some(request.read_lsn.try_into()?),
replica: request.replica,
})
}
}
pub type GetBaseBackupResponseChunk = Bytes;
impl TryFrom<proto::GetBaseBackupResponseChunk> for GetBaseBackupResponseChunk {
type Error = ProtocolError;
fn try_from(pb: proto::GetBaseBackupResponseChunk) -> Result<Self, Self::Error> {
if pb.chunk.is_empty() {
return Err(ProtocolError::Missing("chunk"));
}
Ok(pb.chunk)
}
}
impl TryFrom<GetBaseBackupResponseChunk> for proto::GetBaseBackupResponseChunk {
type Error = ProtocolError;
fn try_from(chunk: GetBaseBackupResponseChunk) -> Result<Self, Self::Error> {
if chunk.is_empty() {
return Err(ProtocolError::Missing("chunk"));
}
Ok(Self { chunk })
}
}
/// Requests the size of a database, as # of bytes. Only valid on shard 0, other shards will error.
#[derive(Clone, Copy, Debug)]
pub struct GetDbSizeRequest {
pub read_lsn: ReadLsn,
pub db_oid: Oid,
}
impl TryFrom<proto::GetDbSizeRequest> for GetDbSizeRequest {
type Error = ProtocolError;
fn try_from(pb: proto::GetDbSizeRequest) -> Result<Self, Self::Error> {
Ok(Self {
read_lsn: pb
.read_lsn
.ok_or(ProtocolError::Missing("read_lsn"))?
.try_into()?,
db_oid: pb.db_oid,
})
}
}
impl TryFrom<GetDbSizeRequest> for proto::GetDbSizeRequest {
type Error = ProtocolError;
fn try_from(request: GetDbSizeRequest) -> Result<Self, Self::Error> {
Ok(Self {
read_lsn: Some(request.read_lsn.try_into()?),
db_oid: request.db_oid,
})
}
}
pub type GetDbSizeResponse = u64;
impl From<proto::GetDbSizeResponse> for GetDbSizeResponse {
fn from(pb: proto::GetDbSizeResponse) -> Self {
pb.num_bytes
}
}
impl From<GetDbSizeResponse> for proto::GetDbSizeResponse {
fn from(num_bytes: GetDbSizeResponse) -> Self {
Self { num_bytes }
}
}
/// Requests one or more pages.
#[derive(Clone, Debug)]
pub struct GetPageRequest {
/// A request ID. Will be included in the response. Should be unique for in-flight requests on
/// the stream.
pub request_id: RequestID,
/// The request class.
pub request_class: GetPageClass,
/// The LSN to read at.
pub read_lsn: ReadLsn,
/// The relation to read from.
pub rel: RelTag,
/// Page numbers to read. Must belong to the remote shard.
///
/// Multiple pages will be executed as a single batch by the Pageserver, amortizing layer access
/// costs and parallelizing them. This may increase the latency of any individual request, but
/// improves the overall latency and throughput of the batch as a whole.
pub block_numbers: SmallVec<[u32; 1]>,
}
impl TryFrom<proto::GetPageRequest> for GetPageRequest {
type Error = ProtocolError;
fn try_from(pb: proto::GetPageRequest) -> Result<Self, Self::Error> {
if pb.block_number.is_empty() {
return Err(ProtocolError::Missing("block_number"));
}
Ok(Self {
request_id: pb.request_id,
request_class: pb.request_class.into(),
read_lsn: pb
.read_lsn
.ok_or(ProtocolError::Missing("read_lsn"))?
.try_into()?,
rel: pb.rel.ok_or(ProtocolError::Missing("rel"))?.try_into()?,
block_numbers: pb.block_number.into(),
})
}
}
impl TryFrom<GetPageRequest> for proto::GetPageRequest {
type Error = ProtocolError;
fn try_from(request: GetPageRequest) -> Result<Self, Self::Error> {
if request.block_numbers.is_empty() {
return Err(ProtocolError::Missing("block_number"));
}
Ok(Self {
request_id: request.request_id,
request_class: request.request_class.into(),
read_lsn: Some(request.read_lsn.try_into()?),
rel: Some(request.rel.into()),
block_number: request.block_numbers.into_vec(),
})
}
}
/// A GetPage request ID.
pub type RequestID = u64;
/// A GetPage request class.
#[derive(Clone, Copy, Debug)]
pub enum GetPageClass {
/// Unknown class. For backwards compatibility: used when an older client version sends a class
/// that a newer server version has removed.
Unknown,
/// A normal request. This is the default.
Normal,
/// A prefetch request. NB: can only be classified on pg < 18.
Prefetch,
/// A background request (e.g. vacuum).
Background,
}
impl From<proto::GetPageClass> for GetPageClass {
fn from(pb: proto::GetPageClass) -> Self {
match pb {
proto::GetPageClass::Unknown => Self::Unknown,
proto::GetPageClass::Normal => Self::Normal,
proto::GetPageClass::Prefetch => Self::Prefetch,
proto::GetPageClass::Background => Self::Background,
}
}
}
impl From<i32> for GetPageClass {
fn from(class: i32) -> Self {
proto::GetPageClass::try_from(class)
.unwrap_or(proto::GetPageClass::Unknown)
.into()
}
}
impl From<GetPageClass> for proto::GetPageClass {
fn from(class: GetPageClass) -> Self {
match class {
GetPageClass::Unknown => Self::Unknown,
GetPageClass::Normal => Self::Normal,
GetPageClass::Prefetch => Self::Prefetch,
GetPageClass::Background => Self::Background,
}
}
}
impl From<GetPageClass> for i32 {
fn from(class: GetPageClass) -> Self {
proto::GetPageClass::from(class).into()
}
}
/// A GetPage response.
///
/// A batch response will contain all of the requested pages. We could eagerly emit individual pages
/// as soon as they are ready, but on a readv() Postgres holds buffer pool locks on all pages in the
/// batch and we'll only return once the entire batch is ready, so no one can make use of the
/// individual pages.
#[derive(Clone, Debug)]
pub struct GetPageResponse {
/// The original request's ID.
pub request_id: RequestID,
/// The response status code.
pub status_code: GetPageStatusCode,
/// A string describing the status, if any.
pub reason: Option<String>,
/// The 8KB page images, in the same order as the request. Empty if status != OK.
pub page_images: SmallVec<[Bytes; 1]>,
}
impl From<proto::GetPageResponse> for GetPageResponse {
fn from(pb: proto::GetPageResponse) -> Self {
Self {
request_id: pb.request_id,
status_code: pb.status_code.into(),
reason: Some(pb.reason).filter(|r| !r.is_empty()),
page_images: pb.page_image.into(),
}
}
}
impl From<GetPageResponse> for proto::GetPageResponse {
fn from(response: GetPageResponse) -> Self {
Self {
request_id: response.request_id,
status_code: response.status_code.into(),
reason: response.reason.unwrap_or_default(),
page_image: response.page_images.into_vec(),
}
}
}
/// A GetPage response status code.
///
/// These are effectively equivalent to gRPC statuses. However, we use a bidirectional stream
/// (potentially shared by many backends), and a gRPC status response would terminate the stream so
/// we send GetPageResponse messages with these codes instead.
#[derive(Clone, Copy, Debug)]
pub enum GetPageStatusCode {
/// Unknown status. For forwards compatibility: used when an older client version receives a new
/// status code from a newer server version.
Unknown,
/// The request was successful.
Ok,
/// The page did not exist. The tenant/timeline/shard has already been validated during stream
/// setup.
NotFound,
/// The request was invalid.
InvalidRequest,
/// The request failed due to an internal server error.
InternalError,
/// The tenant is rate limited. Slow down and retry later.
SlowDown,
}
impl From<proto::GetPageStatusCode> for GetPageStatusCode {
fn from(pb: proto::GetPageStatusCode) -> Self {
match pb {
proto::GetPageStatusCode::Unknown => Self::Unknown,
proto::GetPageStatusCode::Ok => Self::Ok,
proto::GetPageStatusCode::NotFound => Self::NotFound,
proto::GetPageStatusCode::InvalidRequest => Self::InvalidRequest,
proto::GetPageStatusCode::InternalError => Self::InternalError,
proto::GetPageStatusCode::SlowDown => Self::SlowDown,
}
}
}
impl From<i32> for GetPageStatusCode {
fn from(status_code: i32) -> Self {
proto::GetPageStatusCode::try_from(status_code)
.unwrap_or(proto::GetPageStatusCode::Unknown)
.into()
}
}
impl From<GetPageStatusCode> for proto::GetPageStatusCode {
fn from(status_code: GetPageStatusCode) -> Self {
match status_code {
GetPageStatusCode::Unknown => Self::Unknown,
GetPageStatusCode::Ok => Self::Ok,
GetPageStatusCode::NotFound => Self::NotFound,
GetPageStatusCode::InvalidRequest => Self::InvalidRequest,
GetPageStatusCode::InternalError => Self::InternalError,
GetPageStatusCode::SlowDown => Self::SlowDown,
}
}
}
impl From<GetPageStatusCode> for i32 {
fn from(status_code: GetPageStatusCode) -> Self {
proto::GetPageStatusCode::from(status_code).into()
}
}
// Fetches the size of a relation at a given LSN, as # of blocks. Only valid on shard 0, other
// shards will error.
pub struct GetRelSizeRequest {
pub read_lsn: ReadLsn,
pub rel: RelTag,
}
impl TryFrom<proto::GetRelSizeRequest> for GetRelSizeRequest {
type Error = ProtocolError;
fn try_from(proto: proto::GetRelSizeRequest) -> Result<Self, Self::Error> {
Ok(Self {
read_lsn: proto
.read_lsn
.ok_or(ProtocolError::Missing("read_lsn"))?
.try_into()?,
rel: proto.rel.ok_or(ProtocolError::Missing("rel"))?.try_into()?,
})
}
}
impl TryFrom<GetRelSizeRequest> for proto::GetRelSizeRequest {
type Error = ProtocolError;
fn try_from(request: GetRelSizeRequest) -> Result<Self, Self::Error> {
Ok(Self {
read_lsn: Some(request.read_lsn.try_into()?),
rel: Some(request.rel.into()),
})
}
}
pub type GetRelSizeResponse = u32;
impl From<proto::GetRelSizeResponse> for GetRelSizeResponse {
fn from(proto: proto::GetRelSizeResponse) -> Self {
proto.num_blocks
}
}
impl From<GetRelSizeResponse> for proto::GetRelSizeResponse {
fn from(num_blocks: GetRelSizeResponse) -> Self {
Self { num_blocks }
}
}
/// Requests an SLRU segment. Only valid on shard 0, other shards will error.
pub struct GetSlruSegmentRequest {
pub read_lsn: ReadLsn,
pub kind: SlruKind,
pub segno: u32,
}
impl TryFrom<proto::GetSlruSegmentRequest> for GetSlruSegmentRequest {
type Error = ProtocolError;
fn try_from(pb: proto::GetSlruSegmentRequest) -> Result<Self, Self::Error> {
Ok(Self {
read_lsn: pb
.read_lsn
.ok_or(ProtocolError::Missing("read_lsn"))?
.try_into()?,
kind: u8::try_from(pb.kind)
.ok()
.and_then(SlruKind::from_repr)
.ok_or_else(|| ProtocolError::invalid("slru_kind", pb.kind))?,
segno: pb.segno,
})
}
}
impl TryFrom<GetSlruSegmentRequest> for proto::GetSlruSegmentRequest {
type Error = ProtocolError;
fn try_from(request: GetSlruSegmentRequest) -> Result<Self, Self::Error> {
Ok(Self {
read_lsn: Some(request.read_lsn.try_into()?),
kind: request.kind as u32,
segno: request.segno,
})
}
}
pub type GetSlruSegmentResponse = Bytes;
impl TryFrom<proto::GetSlruSegmentResponse> for GetSlruSegmentResponse {
type Error = ProtocolError;
fn try_from(pb: proto::GetSlruSegmentResponse) -> Result<Self, Self::Error> {
if pb.segment.is_empty() {
return Err(ProtocolError::Missing("segment"));
}
Ok(pb.segment)
}
}
impl TryFrom<GetSlruSegmentResponse> for proto::GetSlruSegmentResponse {
type Error = ProtocolError;
fn try_from(segment: GetSlruSegmentResponse) -> Result<Self, Self::Error> {
if segment.is_empty() {
return Err(ProtocolError::Missing("segment"));
}
Ok(Self { segment })
}
}
// SlruKind is defined in pageserver_api::reltag.
pub type SlruKind = pageserver_api::reltag::SlruKind;

View File

@@ -65,9 +65,6 @@ pub(crate) struct Args {
#[clap(long, default_value = "1")]
queue_depth: NonZeroUsize,
#[clap(long)]
only_relnode: Option<u32>,
targets: Option<Vec<TenantTimelineId>>,
}
@@ -209,12 +206,7 @@ async fn main_impl(
for r in partitioning.keys.ranges.iter() {
let mut i = r.start;
while i != r.end {
let mut include = true;
include &= i.is_rel_block_key();
if let Some(only_relnode) = args.only_relnode {
include &= i.is_rel_block_of_rel(only_relnode);
}
if include {
if i.is_rel_block_key() {
filtered.add_key(i);
}
i = i.next();

View File

@@ -144,7 +144,7 @@ where
replica,
ctx,
io_concurrency: IoConcurrency::spawn_from_conf(
timeline.conf.get_vectored_concurrent_io,
timeline.conf,
timeline
.gate
.enter()
@@ -343,7 +343,7 @@ where
// Gather non-relational files from object storage pages.
let slru_partitions = self
.timeline
.get_slru_keyspace(Version::at(self.lsn), self.ctx)
.get_slru_keyspace(Version::Lsn(self.lsn), self.ctx)
.await?
.partition(
self.timeline.get_shard_identity(),
@@ -378,7 +378,7 @@ where
// Otherwise only include init forks of unlogged relations.
let rels = self
.timeline
.list_rels(spcnode, dbnode, Version::at(self.lsn), self.ctx)
.list_rels(spcnode, dbnode, Version::Lsn(self.lsn), self.ctx)
.await?;
for &rel in rels.iter() {
// Send init fork as main fork to provide well formed empty
@@ -517,7 +517,7 @@ where
async fn add_rel(&mut self, src: RelTag, dst: RelTag) -> Result<(), BasebackupError> {
let nblocks = self
.timeline
.get_rel_size(src, Version::at(self.lsn), self.ctx)
.get_rel_size(src, Version::Lsn(self.lsn), self.ctx)
.await?;
// If the relation is empty, create an empty file
@@ -577,7 +577,7 @@ where
let relmap_img = if has_relmap_file {
let img = self
.timeline
.get_relmap_file(spcnode, dbnode, Version::at(self.lsn), self.ctx)
.get_relmap_file(spcnode, dbnode, Version::Lsn(self.lsn), self.ctx)
.await?;
if img.len()
@@ -631,7 +631,7 @@ where
if !has_relmap_file
&& self
.timeline
.list_rels(spcnode, dbnode, Version::at(self.lsn), self.ctx)
.list_rels(spcnode, dbnode, Version::Lsn(self.lsn), self.ctx)
.await?
.is_empty()
{

View File

@@ -1,518 +0,0 @@
use std::{collections::HashMap, sync::Arc};
use async_compression::tokio::write::GzipEncoder;
use camino::{Utf8Path, Utf8PathBuf};
use metrics::core::{AtomicU64, GenericCounter};
use pageserver_api::{config::BasebackupCacheConfig, models::TenantState};
use tokio::{
io::{AsyncWriteExt, BufWriter},
sync::mpsc::{UnboundedReceiver, UnboundedSender},
};
use tokio_util::sync::CancellationToken;
use utils::{
id::{TenantId, TenantTimelineId, TimelineId},
lsn::Lsn,
shard::TenantShardId,
};
use crate::{
basebackup::send_basebackup_tarball,
context::{DownloadBehavior, RequestContext},
metrics::{BASEBACKUP_CACHE_ENTRIES, BASEBACKUP_CACHE_PREPARE, BASEBACKUP_CACHE_READ},
task_mgr::TaskKind,
tenant::{
Timeline,
mgr::{TenantManager, TenantSlot},
},
};
pub struct BasebackupPrepareRequest {
pub tenant_shard_id: TenantShardId,
pub timeline_id: TimelineId,
pub lsn: Lsn,
}
pub type BasebackupPrepareSender = UnboundedSender<BasebackupPrepareRequest>;
pub type BasebackupPrepareReceiver = UnboundedReceiver<BasebackupPrepareRequest>;
type BasebackupRemoveEntrySender = UnboundedSender<Utf8PathBuf>;
type BasebackupRemoveEntryReceiver = UnboundedReceiver<Utf8PathBuf>;
/// BasebackupCache stores cached basebackup archives for timelines on local disk.
///
/// The main purpose of this cache is to speed up the startup process of compute nodes
/// after scaling to zero.
/// Thus, the basebackup is stored only for the latest LSN of the timeline and with
/// fixed set of parameters (gzip=true, full_backup=false, replica=false, prev_lsn=none).
///
/// The cache receives prepare requests through the `BasebackupPrepareSender` channel,
/// generates a basebackup from the timeline in the background, and stores it on disk.
///
/// Basebackup requests are pretty rare. We expect ~thousands of entries in the cache
/// and ~1 RPS for get requests.
pub struct BasebackupCache {
data_dir: Utf8PathBuf,
config: BasebackupCacheConfig,
tenant_manager: Arc<TenantManager>,
remove_entry_sender: BasebackupRemoveEntrySender,
entries: std::sync::Mutex<HashMap<TenantTimelineId, Lsn>>,
cancel: CancellationToken,
read_hit_count: GenericCounter<AtomicU64>,
read_miss_count: GenericCounter<AtomicU64>,
read_err_count: GenericCounter<AtomicU64>,
prepare_ok_count: GenericCounter<AtomicU64>,
prepare_skip_count: GenericCounter<AtomicU64>,
prepare_err_count: GenericCounter<AtomicU64>,
}
impl BasebackupCache {
/// Creates a BasebackupCache and spawns the background task.
/// The initialization of the cache is performed in the background and does not
/// block the caller. The cache will return `None` for any get requests until
/// initialization is complete.
pub fn spawn(
runtime_handle: &tokio::runtime::Handle,
data_dir: Utf8PathBuf,
config: Option<BasebackupCacheConfig>,
prepare_receiver: BasebackupPrepareReceiver,
tenant_manager: Arc<TenantManager>,
cancel: CancellationToken,
) -> Arc<Self> {
let (remove_entry_sender, remove_entry_receiver) = tokio::sync::mpsc::unbounded_channel();
let enabled = config.is_some();
let cache = Arc::new(BasebackupCache {
data_dir,
config: config.unwrap_or_default(),
tenant_manager,
remove_entry_sender,
entries: std::sync::Mutex::new(HashMap::new()),
cancel,
read_hit_count: BASEBACKUP_CACHE_READ.with_label_values(&["hit"]),
read_miss_count: BASEBACKUP_CACHE_READ.with_label_values(&["miss"]),
read_err_count: BASEBACKUP_CACHE_READ.with_label_values(&["error"]),
prepare_ok_count: BASEBACKUP_CACHE_PREPARE.with_label_values(&["ok"]),
prepare_skip_count: BASEBACKUP_CACHE_PREPARE.with_label_values(&["skip"]),
prepare_err_count: BASEBACKUP_CACHE_PREPARE.with_label_values(&["error"]),
});
if enabled {
runtime_handle.spawn(
cache
.clone()
.background(prepare_receiver, remove_entry_receiver),
);
}
cache
}
/// Gets a basebackup entry from the cache.
/// If the entry is found, opens a file with the basebackup archive and returns it.
/// The open file descriptor will prevent the file system from deleting the file
/// even if the entry is removed from the cache in the background.
pub async fn get(
&self,
tenant_id: TenantId,
timeline_id: TimelineId,
lsn: Lsn,
) -> Option<tokio::fs::File> {
// Fast path. Check if the entry exists using the in-memory state.
let tti = TenantTimelineId::new(tenant_id, timeline_id);
if self.entries.lock().unwrap().get(&tti) != Some(&lsn) {
self.read_miss_count.inc();
return None;
}
let path = self.entry_path(tenant_id, timeline_id, lsn);
match tokio::fs::File::open(path).await {
Ok(file) => {
self.read_hit_count.inc();
Some(file)
}
Err(e) => {
if e.kind() == std::io::ErrorKind::NotFound {
// We may end up here if the basebackup was concurrently removed by the cleanup task.
self.read_miss_count.inc();
} else {
self.read_err_count.inc();
tracing::warn!("Unexpected error opening basebackup cache file: {:?}", e);
}
None
}
}
}
// Private methods.
fn entry_filename(tenant_id: TenantId, timeline_id: TimelineId, lsn: Lsn) -> String {
// The default format for LSN is 0/ABCDEF.
// The backslash is not filename friendly, so serialize it as plain hex.
let lsn = lsn.0;
format!("basebackup_{tenant_id}_{timeline_id}_{lsn:016X}.tar.gz")
}
fn entry_path(&self, tenant_id: TenantId, timeline_id: TimelineId, lsn: Lsn) -> Utf8PathBuf {
self.data_dir
.join(Self::entry_filename(tenant_id, timeline_id, lsn))
}
fn entry_tmp_path(
&self,
tenant_id: TenantId,
timeline_id: TimelineId,
lsn: Lsn,
) -> Utf8PathBuf {
self.data_dir
.join("tmp")
.join(Self::entry_filename(tenant_id, timeline_id, lsn))
}
fn parse_entry_filename(filename: &str) -> Option<(TenantId, TimelineId, Lsn)> {
let parts: Vec<&str> = filename
.strip_prefix("basebackup_")?
.strip_suffix(".tar.gz")?
.split('_')
.collect();
if parts.len() != 3 {
return None;
}
let tenant_id = parts[0].parse::<TenantId>().ok()?;
let timeline_id = parts[1].parse::<TimelineId>().ok()?;
let lsn = Lsn(u64::from_str_radix(parts[2], 16).ok()?);
Some((tenant_id, timeline_id, lsn))
}
async fn cleanup(&self) -> anyhow::Result<()> {
// Cleanup tmp directory.
let tmp_dir = self.data_dir.join("tmp");
let mut tmp_dir = tokio::fs::read_dir(&tmp_dir).await?;
while let Some(dir_entry) = tmp_dir.next_entry().await? {
if let Err(e) = tokio::fs::remove_file(dir_entry.path()).await {
tracing::warn!("Failed to remove basebackup cache tmp file: {:#}", e);
}
}
// Remove outdated entries.
let entries_old = self.entries.lock().unwrap().clone();
let mut entries_new = HashMap::new();
for (tenant_shard_id, tenant_slot) in self.tenant_manager.list() {
if !tenant_shard_id.is_shard_zero() {
continue;
}
let TenantSlot::Attached(tenant) = tenant_slot else {
continue;
};
let tenant_id = tenant_shard_id.tenant_id;
for timeline in tenant.list_timelines() {
let tti = TenantTimelineId::new(tenant_id, timeline.timeline_id);
if let Some(&entry_lsn) = entries_old.get(&tti) {
if timeline.get_last_record_lsn() <= entry_lsn {
entries_new.insert(tti, entry_lsn);
}
}
}
}
for (&tti, &lsn) in entries_old.iter() {
if !entries_new.contains_key(&tti) {
self.remove_entry_sender
.send(self.entry_path(tti.tenant_id, tti.timeline_id, lsn))
.unwrap();
}
}
BASEBACKUP_CACHE_ENTRIES.set(entries_new.len() as i64);
*self.entries.lock().unwrap() = entries_new;
Ok(())
}
async fn on_startup(&self) -> anyhow::Result<()> {
// Create data_dir and tmp directory if they do not exist.
tokio::fs::create_dir_all(&self.data_dir.join("tmp"))
.await
.map_err(|e| {
anyhow::anyhow!(
"Failed to create basebackup cache data_dir {:?}: {:?}",
self.data_dir,
e
)
})?;
// Read existing entries from the data_dir and add them to in-memory state.
let mut entries = HashMap::new();
let mut dir = tokio::fs::read_dir(&self.data_dir).await?;
while let Some(dir_entry) = dir.next_entry().await? {
let filename = dir_entry.file_name();
if filename == "tmp" {
// Skip the tmp directory.
continue;
}
let parsed = Self::parse_entry_filename(filename.to_string_lossy().as_ref());
let Some((tenant_id, timeline_id, lsn)) = parsed else {
tracing::warn!("Invalid basebackup cache file name: {:?}", filename);
continue;
};
let tti = TenantTimelineId::new(tenant_id, timeline_id);
use std::collections::hash_map::Entry::*;
match entries.entry(tti) {
Occupied(mut entry) => {
let entry_lsn = *entry.get();
// Leave only the latest entry, remove the old one.
if lsn < entry_lsn {
self.remove_entry_sender.send(self.entry_path(
tenant_id,
timeline_id,
lsn,
))?;
} else if lsn > entry_lsn {
self.remove_entry_sender.send(self.entry_path(
tenant_id,
timeline_id,
entry_lsn,
))?;
entry.insert(lsn);
} else {
// Two different filenames parsed to the same timline_id and LSN.
// Should never happen.
return Err(anyhow::anyhow!(
"Duplicate basebackup cache entry with the same LSN: {:?}",
filename
));
}
}
Vacant(entry) => {
entry.insert(lsn);
}
}
}
BASEBACKUP_CACHE_ENTRIES.set(entries.len() as i64);
*self.entries.lock().unwrap() = entries;
Ok(())
}
async fn background(
self: Arc<Self>,
mut prepare_receiver: BasebackupPrepareReceiver,
mut remove_entry_receiver: BasebackupRemoveEntryReceiver,
) {
// Panic in the background is a safe fallback.
// It will drop receivers and the cache will be effectively disabled.
self.on_startup()
.await
.expect("Failed to initialize basebackup cache");
let mut cleanup_ticker = tokio::time::interval(self.config.cleanup_period);
cleanup_ticker.set_missed_tick_behavior(tokio::time::MissedTickBehavior::Skip);
loop {
tokio::select! {
Some(req) = prepare_receiver.recv() => {
if let Err(err) = self.prepare_basebackup(
req.tenant_shard_id,
req.timeline_id,
req.lsn,
).await {
tracing::info!("Failed to prepare basebackup: {:#}", err);
self.prepare_err_count.inc();
continue;
}
}
Some(req) = remove_entry_receiver.recv() => {
if let Err(e) = tokio::fs::remove_file(req).await {
tracing::warn!("Failed to remove basebackup cache file: {:#}", e);
}
}
_ = cleanup_ticker.tick() => {
self.cleanup().await.unwrap_or_else(|e| {
tracing::warn!("Failed to clean up basebackup cache: {:#}", e);
});
}
_ = self.cancel.cancelled() => {
tracing::info!("BasebackupCache background task cancelled");
break;
}
}
}
}
/// Prepare a basebackup for the given timeline.
///
/// If the basebackup already exists with a higher LSN or the timeline already
/// has a higher last_record_lsn, skip the preparation.
///
/// The basebackup is prepared in a temporary directory and then moved to the final
/// location to make the operation atomic.
async fn prepare_basebackup(
&self,
tenant_shard_id: TenantShardId,
timeline_id: TimelineId,
req_lsn: Lsn,
) -> anyhow::Result<()> {
tracing::info!(
tenant_id = %tenant_shard_id.tenant_id,
%timeline_id,
%req_lsn,
"Preparing basebackup for timeline",
);
let tti = TenantTimelineId::new(tenant_shard_id.tenant_id, timeline_id);
{
let entries = self.entries.lock().unwrap();
if let Some(&entry_lsn) = entries.get(&tti) {
if entry_lsn >= req_lsn {
tracing::info!(
%timeline_id,
%req_lsn,
%entry_lsn,
"Basebackup entry already exists for timeline with higher LSN, skipping basebackup",
);
self.prepare_skip_count.inc();
return Ok(());
}
}
if entries.len() as i64 >= self.config.max_size_entries {
tracing::info!(
%timeline_id,
%req_lsn,
"Basebackup cache is full, skipping basebackup",
);
self.prepare_skip_count.inc();
return Ok(());
}
}
let tenant = self
.tenant_manager
.get_attached_tenant_shard(tenant_shard_id)?;
let tenant_state = tenant.current_state();
if tenant_state != TenantState::Active {
anyhow::bail!(
"Tenant {} is not active, current state: {:?}",
tenant_shard_id.tenant_id,
tenant_state
)
}
let timeline = tenant.get_timeline(timeline_id, true)?;
let last_record_lsn = timeline.get_last_record_lsn();
if last_record_lsn > req_lsn {
tracing::info!(
%timeline_id,
%req_lsn,
%last_record_lsn,
"Timeline has a higher LSN than the requested one, skipping basebackup",
);
self.prepare_skip_count.inc();
return Ok(());
}
let entry_tmp_path = self.entry_tmp_path(tenant_shard_id.tenant_id, timeline_id, req_lsn);
let res = self
.prepare_basebackup_tmp(&entry_tmp_path, &timeline, req_lsn)
.await;
if let Err(err) = res {
tracing::info!("Failed to prepare basebackup tmp file: {:#}", err);
// Try to clean up tmp file. If we fail, the background clean up task will take care of it.
match tokio::fs::remove_file(&entry_tmp_path).await {
Ok(_) => {}
Err(e) if e.kind() == std::io::ErrorKind::NotFound => {}
Err(e) => {
tracing::info!("Failed to remove basebackup tmp file: {:?}", e);
}
}
return Err(err);
}
// Move the tmp file to the final location atomically.
let entry_path = self.entry_path(tenant_shard_id.tenant_id, timeline_id, req_lsn);
tokio::fs::rename(&entry_tmp_path, &entry_path).await?;
let mut entries = self.entries.lock().unwrap();
if let Some(old_lsn) = entries.insert(tti, req_lsn) {
// Remove the old entry if it exists.
self.remove_entry_sender
.send(self.entry_path(tenant_shard_id.tenant_id, timeline_id, old_lsn))
.unwrap();
}
BASEBACKUP_CACHE_ENTRIES.set(entries.len() as i64);
self.prepare_ok_count.inc();
Ok(())
}
/// Prepares a basebackup in a temporary file.
async fn prepare_basebackup_tmp(
&self,
emptry_tmp_path: &Utf8Path,
timeline: &Arc<Timeline>,
req_lsn: Lsn,
) -> anyhow::Result<()> {
let ctx = RequestContext::new(TaskKind::BasebackupCache, DownloadBehavior::Download);
let ctx = ctx.with_scope_timeline(timeline);
let file = tokio::fs::File::create(emptry_tmp_path).await?;
let mut writer = BufWriter::new(file);
let mut encoder = GzipEncoder::with_quality(
&mut writer,
// Level::Best because compression is not on the hot path of basebackup requests.
// The decompression is almost not affected by the compression level.
async_compression::Level::Best,
);
// We may receive a request before the WAL record is applied to the timeline.
// Wait for the requested LSN to be applied.
timeline
.wait_lsn(
req_lsn,
crate::tenant::timeline::WaitLsnWaiter::BaseBackupCache,
crate::tenant::timeline::WaitLsnTimeout::Default,
&ctx,
)
.await?;
send_basebackup_tarball(
&mut encoder,
timeline,
Some(req_lsn),
None,
false,
false,
&ctx,
)
.await?;
encoder.shutdown().await?;
writer.flush().await?;
writer.into_inner().sync_all().await?;
Ok(())
}
}

View File

@@ -16,12 +16,10 @@ use http_utils::tls_certs::ReloadingCertificateResolver;
use metrics::launch_timestamp::{LaunchTimestamp, set_launch_timestamp_metric};
use metrics::set_build_info_metric;
use nix::sys::socket::{setsockopt, sockopt};
use pageserver::basebackup_cache::BasebackupCache;
use pageserver::config::{PageServerConf, PageserverIdentity, ignored_fields};
use pageserver::controller_upcall_client::StorageControllerUpcallClient;
use pageserver::deletion_queue::DeletionQueue;
use pageserver::disk_usage_eviction_task::{self, launch_disk_usage_global_eviction_task};
use pageserver::feature_resolver::FeatureResolver;
use pageserver::metrics::{STARTUP_DURATION, STARTUP_IS_LOADING};
use pageserver::task_mgr::{
BACKGROUND_RUNTIME, COMPUTE_REQUEST_RUNTIME, MGMT_REQUEST_RUNTIME, WALRECEIVER_RUNTIME,
@@ -389,30 +387,23 @@ fn start_pageserver(
// We need to release the lock file only when the process exits.
std::mem::forget(lock_file);
// Bind the HTTP, libpq, and gRPC ports early, to error out if they are
// already in use.
info!(
"Starting pageserver http handler on {} with auth {:#?}",
conf.listen_http_addr, conf.http_auth_type
);
let http_listener = tcp_listener::bind(&conf.listen_http_addr)?;
// Bind the HTTP and libpq ports early, so that if they are in use by some other
// process, we error out early.
let http_addr = &conf.listen_http_addr;
info!("Starting pageserver http handler on {http_addr}");
let http_listener = tcp_listener::bind(http_addr)?;
let https_listener = match conf.listen_https_addr.as_ref() {
Some(https_addr) => {
info!(
"Starting pageserver https handler on {https_addr} with auth {:#?}",
conf.http_auth_type
);
info!("Starting pageserver https handler on {https_addr}");
Some(tcp_listener::bind(https_addr)?)
}
None => None,
};
info!(
"Starting pageserver pg protocol handler on {} with auth {:#?}",
conf.listen_pg_addr, conf.pg_auth_type,
);
let pageserver_listener = tcp_listener::bind(&conf.listen_pg_addr)?;
let pg_addr = &conf.listen_pg_addr;
info!("Starting pageserver pg protocol handler on {pg_addr}");
let pageserver_listener = tcp_listener::bind(pg_addr)?;
// Enable SO_KEEPALIVE on the socket, to detect dead connections faster.
// These are configured via net.ipv4.tcp_keepalive_* sysctls.
@@ -421,15 +412,6 @@ fn start_pageserver(
// support enabling keepalives while using the default OS sysctls.
setsockopt(&pageserver_listener, sockopt::KeepAlive, &true)?;
let mut grpc_listener = None;
if let Some(grpc_addr) = &conf.listen_grpc_addr {
info!(
"Starting pageserver gRPC handler on {grpc_addr} with auth {:#?}",
conf.grpc_auth_type
);
grpc_listener = Some(tcp_listener::bind(grpc_addr).map_err(|e| anyhow!("{e}"))?);
}
// Launch broker client
// The storage_broker::connect call needs to happen inside a tokio runtime thread.
let broker_client = WALRECEIVER_RUNTIME
@@ -457,8 +439,7 @@ fn start_pageserver(
// Initialize authentication for incoming connections
let http_auth;
let pg_auth;
let grpc_auth;
if [conf.http_auth_type, conf.pg_auth_type, conf.grpc_auth_type].contains(&AuthType::NeonJWT) {
if conf.http_auth_type == AuthType::NeonJWT || conf.pg_auth_type == AuthType::NeonJWT {
// unwrap is ok because check is performed when creating config, so path is set and exists
let key_path = conf.auth_validation_public_key_path.as_ref().unwrap();
info!("Loading public key(s) for verifying JWT tokens from {key_path:?}");
@@ -466,23 +447,20 @@ fn start_pageserver(
let jwt_auth = JwtAuth::from_key_path(key_path)?;
let auth: Arc<SwappableJwtAuth> = Arc::new(SwappableJwtAuth::new(jwt_auth));
http_auth = match conf.http_auth_type {
http_auth = match &conf.http_auth_type {
AuthType::Trust => None,
AuthType::NeonJWT => Some(auth.clone()),
};
pg_auth = match conf.pg_auth_type {
AuthType::Trust => None,
AuthType::NeonJWT => Some(auth.clone()),
};
grpc_auth = match conf.grpc_auth_type {
pg_auth = match &conf.pg_auth_type {
AuthType::Trust => None,
AuthType::NeonJWT => Some(auth),
};
} else {
http_auth = None;
pg_auth = None;
grpc_auth = None;
}
info!("Using auth for http API: {:#?}", conf.http_auth_type);
info!("Using auth for pg connections: {:#?}", conf.pg_auth_type);
let tls_server_config = if conf.listen_https_addr.is_some() || conf.enable_tls_page_service_api
{
@@ -523,12 +501,6 @@ fn start_pageserver(
// Set up remote storage client
let remote_storage = BACKGROUND_RUNTIME.block_on(create_remote_storage_client(conf))?;
let feature_resolver = create_feature_resolver(
conf,
shutdown_pageserver.clone(),
BACKGROUND_RUNTIME.handle(),
)?;
// Set up deletion queue
let (deletion_queue, deletion_workers) = DeletionQueue::new(
remote_storage.clone(),
@@ -569,8 +541,6 @@ fn start_pageserver(
pageserver::l0_flush::L0FlushGlobalState::new(conf.l0_flush.clone());
// Scan the local 'tenants/' directory and start loading the tenants
let (basebackup_prepare_sender, basebackup_prepare_receiver) =
tokio::sync::mpsc::unbounded_channel();
let deletion_queue_client = deletion_queue.new_client();
let background_purges = mgr::BackgroundPurges::default();
let tenant_manager = BACKGROUND_RUNTIME.block_on(mgr::init_tenant_mgr(
@@ -581,23 +551,12 @@ fn start_pageserver(
remote_storage: remote_storage.clone(),
deletion_queue_client,
l0_flush_global_state,
basebackup_prepare_sender,
feature_resolver,
},
order,
shutdown_pageserver.clone(),
))?;
let tenant_manager = Arc::new(tenant_manager);
let basebackup_cache = BasebackupCache::spawn(
BACKGROUND_RUNTIME.handle(),
conf.basebackup_cache_dir(),
conf.basebackup_cache_config.clone(),
basebackup_prepare_receiver,
Arc::clone(&tenant_manager),
shutdown_pageserver.child_token(),
);
BACKGROUND_RUNTIME.spawn({
let shutdown_pageserver = shutdown_pageserver.clone();
let drive_init = async move {
@@ -804,27 +763,8 @@ fn start_pageserver(
} else {
None
},
basebackup_cache.clone(),
);
// Spawn a Pageserver gRPC server task. It will spawn separate tasks for
// each stream/request.
//
// TODO: this uses a separate Tokio runtime for the page service. If we want
// other gRPC services, they will need their own port and runtime. Is this
// necessary?
let mut page_service_grpc = None;
if let Some(grpc_listener) = grpc_listener {
page_service_grpc = Some(page_service::spawn_grpc(
conf,
tenant_manager.clone(),
grpc_auth,
otel_guard.as_ref().map(|g| g.dispatch.clone()),
grpc_listener,
basebackup_cache,
)?);
}
// All started up! Now just sit and wait for shutdown signal.
BACKGROUND_RUNTIME.block_on(async move {
let signal_token = CancellationToken::new();
@@ -843,7 +783,6 @@ fn start_pageserver(
http_endpoint_listener,
https_endpoint_listener,
page_service,
page_service_grpc,
consumption_metrics_tasks,
disk_usage_eviction_task,
&tenant_manager,
@@ -857,14 +796,6 @@ fn start_pageserver(
})
}
fn create_feature_resolver(
conf: &'static PageServerConf,
shutdown_pageserver: CancellationToken,
handle: &tokio::runtime::Handle,
) -> anyhow::Result<FeatureResolver> {
FeatureResolver::spawn(conf, shutdown_pageserver, handle)
}
async fn create_remote_storage_client(
conf: &'static PageServerConf,
) -> anyhow::Result<GenericRemoteStorage> {

View File

@@ -14,7 +14,7 @@ use std::time::Duration;
use anyhow::{Context, bail, ensure};
use camino::{Utf8Path, Utf8PathBuf};
use once_cell::sync::OnceCell;
use pageserver_api::config::{DiskUsageEvictionTaskConfig, MaxVectoredReadBytes, PostHogConfig};
use pageserver_api::config::{DiskUsageEvictionTaskConfig, MaxVectoredReadBytes};
use pageserver_api::models::ImageCompressionAlgorithm;
use pageserver_api::shard::TenantShardId;
use pem::Pem;
@@ -58,16 +58,11 @@ pub struct PageServerConf {
pub listen_http_addr: String,
/// Example: 127.0.0.1:9899
pub listen_https_addr: Option<String>,
/// If set, expose a gRPC API on this address.
/// Example: 127.0.0.1:51051
///
/// EXPERIMENTAL: this protocol is unstable and under active development.
pub listen_grpc_addr: Option<String>,
/// Path to a file with certificate's private key for https and gRPC API.
/// Path to a file with certificate's private key for https API.
/// Default: server.key
pub ssl_key_file: Utf8PathBuf,
/// Path to a file with a X509 certificate for https and gRPC API.
/// Path to a file with a X509 certificate for https API.
/// Default: server.crt
pub ssl_cert_file: Utf8PathBuf,
/// Period to reload certificate and private key from files.
@@ -105,8 +100,6 @@ pub struct PageServerConf {
pub http_auth_type: AuthType,
/// authentication method for libpq connections from compute
pub pg_auth_type: AuthType,
/// authentication method for gRPC connections from compute
pub grpc_auth_type: AuthType,
/// Path to a file or directory containing public key(s) for verifying JWT tokens.
/// Used for both mgmt and compute auth, if enabled.
pub auth_validation_public_key_path: Option<Utf8PathBuf>,
@@ -238,12 +231,7 @@ pub struct PageServerConf {
/// This is insecure and should only be used in development environments.
pub dev_mode: bool,
/// PostHog integration config.
pub posthog_config: Option<PostHogConfig>,
pub timeline_import_config: pageserver_api::config::TimelineImportConfig,
pub basebackup_cache_config: Option<pageserver_api::config::BasebackupCacheConfig>,
}
/// Token for authentication to safekeepers
@@ -273,10 +261,6 @@ impl PageServerConf {
self.workdir.join("metadata.json")
}
pub fn basebackup_cache_dir(&self) -> Utf8PathBuf {
self.workdir.join("basebackup_cache")
}
pub fn deletion_list_path(&self, sequence: u64) -> Utf8PathBuf {
// Encode a version in the filename, so that if we ever switch away from JSON we can
// increment this.
@@ -365,7 +349,6 @@ impl PageServerConf {
listen_pg_addr,
listen_http_addr,
listen_https_addr,
listen_grpc_addr,
ssl_key_file,
ssl_cert_file,
ssl_cert_reload_period,
@@ -380,7 +363,6 @@ impl PageServerConf {
pg_distrib_dir,
http_auth_type,
pg_auth_type,
grpc_auth_type,
auth_validation_public_key_path,
remote_storage,
broker_endpoint,
@@ -424,9 +406,7 @@ impl PageServerConf {
tracing,
enable_tls_page_service_api,
dev_mode,
posthog_config,
timeline_import_config,
basebackup_cache_config,
} = config_toml;
let mut conf = PageServerConf {
@@ -436,7 +416,6 @@ impl PageServerConf {
listen_pg_addr,
listen_http_addr,
listen_https_addr,
listen_grpc_addr,
ssl_key_file,
ssl_cert_file,
ssl_cert_reload_period,
@@ -449,7 +428,6 @@ impl PageServerConf {
max_file_descriptors,
http_auth_type,
pg_auth_type,
grpc_auth_type,
auth_validation_public_key_path,
remote_storage_config: remote_storage,
broker_endpoint,
@@ -483,7 +461,6 @@ impl PageServerConf {
enable_tls_page_service_api,
dev_mode,
timeline_import_config,
basebackup_cache_config,
// ------------------------------------------------------------
// fields that require additional validation or custom handling
@@ -540,16 +517,13 @@ impl PageServerConf {
}
None => Vec::new(),
},
posthog_config,
};
// ------------------------------------------------------------
// custom validation code that covers more than one field in isolation
// ------------------------------------------------------------
if [conf.http_auth_type, conf.pg_auth_type, conf.grpc_auth_type]
.contains(&AuthType::NeonJWT)
{
if conf.http_auth_type == AuthType::NeonJWT || conf.pg_auth_type == AuthType::NeonJWT {
let auth_validation_public_key_path = conf
.auth_validation_public_key_path
.get_or_insert_with(|| workdir.join("auth_public_key.pem"));
@@ -570,23 +544,6 @@ impl PageServerConf {
ratio.numerator, ratio.denominator
)
);
let url = Url::parse(&tracing_config.export_config.endpoint)
.map_err(anyhow::Error::msg)
.with_context(|| {
format!(
"tracing endpoint URL is invalid : {}",
tracing_config.export_config.endpoint
)
})?;
ensure!(
url.scheme() == "http" || url.scheme() == "https",
format!(
"tracing endpoint URL must start with http:// or https://: {}",
tracing_config.export_config.endpoint
)
);
}
IndexEntry::validate_checkpoint_distance(conf.default_tenant_conf.checkpoint_distance)
@@ -703,25 +660,4 @@ mod tests {
PageServerConf::parse_and_validate(NodeId(0), config_toml, &workdir)
.expect("parse_and_validate");
}
#[test]
fn test_config_tracing_endpoint_is_invalid() {
let input = r#"
control_plane_api = "http://localhost:6666"
[tracing]
sampling_ratio = { numerator = 1, denominator = 0 }
[tracing.export_config]
endpoint = "localhost:4317"
protocol = "http-binary"
timeout = "1ms"
"#;
let config_toml = toml_edit::de::from_str::<pageserver_api::config::ConfigToml>(input)
.expect("config has valid fields");
let workdir = Utf8PathBuf::from("/nonexistent");
PageServerConf::parse_and_validate(NodeId(0), config_toml, &workdir)
.expect_err("parse_and_validate should fail for endpoint without scheme");
}
}

View File

@@ -18,25 +18,12 @@ use crate::tenant::timeline::logical_size::CurrentLogicalSize;
// management.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, serde::Serialize, serde::Deserialize)]
pub(super) enum Name {
/// Timeline last_record_lsn, absolute.
/// Timeline last_record_lsn, absolute
#[serde(rename = "written_size")]
WrittenSize,
/// Timeline last_record_lsn, incremental
#[serde(rename = "written_data_bytes_delta")]
WrittenSizeDelta,
/// Written bytes only on this timeline (not including ancestors):
/// written_size - ancestor_lsn
///
/// On the root branch, this is equivalent to `written_size`.
#[serde(rename = "written_size_since_parent")]
WrittenSizeSinceParent,
/// PITR history size only on this timeline (not including ancestors):
/// last_record_lsn - max(pitr_cutoff, ancestor_lsn).
///
/// On the root branch, this is its entire PITR history size. Not emitted if GC hasn't computed
/// the PITR cutoff yet. 0 if PITR is disabled.
#[serde(rename = "pitr_history_size_since_parent")]
PitrHistorySizeSinceParent,
/// Timeline logical size
#[serde(rename = "timeline_logical_size")]
LogicalSize,
@@ -170,32 +157,6 @@ impl MetricsKey {
.incremental_values()
}
/// `written_size` - `ancestor_lsn`.
const fn written_size_since_parent(
tenant_id: TenantId,
timeline_id: TimelineId,
) -> AbsoluteValueFactory {
MetricsKey {
tenant_id,
timeline_id: Some(timeline_id),
metric: Name::WrittenSizeSinceParent,
}
.absolute_values()
}
/// `written_size` - max(`pitr_cutoff`, `ancestor_lsn`).
const fn pitr_history_size_since_parent(
tenant_id: TenantId,
timeline_id: TimelineId,
) -> AbsoluteValueFactory {
MetricsKey {
tenant_id,
timeline_id: Some(timeline_id),
metric: Name::PitrHistorySizeSinceParent,
}
.absolute_values()
}
/// Exact [`Timeline::get_current_logical_size`].
///
/// [`Timeline::get_current_logical_size`]: crate::tenant::Timeline::get_current_logical_size
@@ -373,13 +334,7 @@ impl TenantSnapshot {
struct TimelineSnapshot {
loaded_at: (Lsn, SystemTime),
last_record_lsn: Lsn,
ancestor_lsn: Lsn,
current_exact_logical_size: Option<u64>,
/// Whether PITR is enabled (pitr_interval > 0).
pitr_enabled: bool,
/// The PITR cutoff LSN. None if not yet initialized. If PITR is disabled, this is approximately
/// Some(last_record_lsn), but may lag behind it since it's computed periodically.
pitr_cutoff: Option<Lsn>,
}
impl TimelineSnapshot {
@@ -399,9 +354,6 @@ impl TimelineSnapshot {
} else {
let loaded_at = t.loaded_at;
let last_record_lsn = t.get_last_record_lsn();
let ancestor_lsn = t.get_ancestor_lsn();
let pitr_enabled = !t.get_pitr_interval().is_zero();
let pitr_cutoff = t.gc_info.read().unwrap().cutoffs.time;
let current_exact_logical_size = {
let span = tracing::info_span!("collect_metrics_iteration", tenant_id = %t.tenant_shard_id.tenant_id, timeline_id = %t.timeline_id);
@@ -421,10 +373,7 @@ impl TimelineSnapshot {
Ok(Some(TimelineSnapshot {
loaded_at,
last_record_lsn,
ancestor_lsn,
current_exact_logical_size,
pitr_enabled,
pitr_cutoff,
}))
}
}
@@ -475,8 +424,6 @@ impl TimelineSnapshot {
let up_to = now;
let written_size_last = written_size_now.value.max(prev.1); // don't regress
if let Some(delta) = written_size_now.value.checked_sub(prev.1) {
let key_value = written_size_delta_key.from_until(prev.0, up_to, delta);
// written_size_delta
@@ -494,27 +441,6 @@ impl TimelineSnapshot {
});
}
// Compute the branch-local written size.
let written_size_since_parent_key =
MetricsKey::written_size_since_parent(tenant_id, timeline_id);
metrics.push(
written_size_since_parent_key
.at(now, written_size_last.saturating_sub(self.ancestor_lsn.0)),
);
// Compute the branch-local PITR history size. Not emitted if GC hasn't yet computed the
// PITR cutoff. 0 if PITR is disabled.
let pitr_history_size_since_parent_key =
MetricsKey::pitr_history_size_since_parent(tenant_id, timeline_id);
if !self.pitr_enabled {
metrics.push(pitr_history_size_since_parent_key.at(now, 0));
} else if let Some(pitr_cutoff) = self.pitr_cutoff {
metrics.push(pitr_history_size_since_parent_key.at(
now,
written_size_last.saturating_sub(pitr_cutoff.max(self.ancestor_lsn).0),
));
}
{
let factory = MetricsKey::timeline_logical_size(tenant_id, timeline_id);
let current_or_previous = self

View File

@@ -12,17 +12,12 @@ fn startup_collected_timeline_metrics_before_advancing() {
let cache = HashMap::new();
let initdb_lsn = Lsn(0x10000);
let pitr_cutoff = Lsn(0x11000);
let disk_consistent_lsn = Lsn(initdb_lsn.0 * 2);
let logical_size = 0x42000;
let snap = TimelineSnapshot {
loaded_at: (disk_consistent_lsn, SystemTime::now()),
last_record_lsn: disk_consistent_lsn,
ancestor_lsn: Lsn(0),
current_exact_logical_size: Some(logical_size),
pitr_enabled: true,
pitr_cutoff: Some(pitr_cutoff),
current_exact_logical_size: Some(0x42000),
};
let now = DateTime::<Utc>::from(SystemTime::now());
@@ -38,11 +33,7 @@ fn startup_collected_timeline_metrics_before_advancing() {
0
),
MetricsKey::written_size(tenant_id, timeline_id).at(now, disk_consistent_lsn.0),
MetricsKey::written_size_since_parent(tenant_id, timeline_id)
.at(now, disk_consistent_lsn.0),
MetricsKey::pitr_history_size_since_parent(tenant_id, timeline_id)
.at(now, disk_consistent_lsn.0 - pitr_cutoff.0),
MetricsKey::timeline_logical_size(tenant_id, timeline_id).at(now, logical_size)
MetricsKey::timeline_logical_size(tenant_id, timeline_id).at(now, 0x42000)
]
);
}
@@ -58,9 +49,7 @@ fn startup_collected_timeline_metrics_second_round() {
let before = DateTime::<Utc>::from(before);
let initdb_lsn = Lsn(0x10000);
let pitr_cutoff = Lsn(0x11000);
let disk_consistent_lsn = Lsn(initdb_lsn.0 * 2);
let logical_size = 0x42000;
let mut metrics = Vec::new();
let cache = HashMap::from([MetricsKey::written_size(tenant_id, timeline_id)
@@ -70,10 +59,7 @@ fn startup_collected_timeline_metrics_second_round() {
let snap = TimelineSnapshot {
loaded_at: (disk_consistent_lsn, init),
last_record_lsn: disk_consistent_lsn,
ancestor_lsn: Lsn(0),
current_exact_logical_size: Some(logical_size),
pitr_enabled: true,
pitr_cutoff: Some(pitr_cutoff),
current_exact_logical_size: Some(0x42000),
};
snap.to_metrics(tenant_id, timeline_id, now, &mut metrics, &cache);
@@ -83,11 +69,7 @@ fn startup_collected_timeline_metrics_second_round() {
&[
MetricsKey::written_size_delta(tenant_id, timeline_id).from_until(before, now, 0),
MetricsKey::written_size(tenant_id, timeline_id).at(now, disk_consistent_lsn.0),
MetricsKey::written_size_since_parent(tenant_id, timeline_id)
.at(now, disk_consistent_lsn.0),
MetricsKey::pitr_history_size_since_parent(tenant_id, timeline_id)
.at(now, disk_consistent_lsn.0 - pitr_cutoff.0),
MetricsKey::timeline_logical_size(tenant_id, timeline_id).at(now, logical_size)
MetricsKey::timeline_logical_size(tenant_id, timeline_id).at(now, 0x42000)
]
);
}
@@ -104,9 +86,7 @@ fn startup_collected_timeline_metrics_nth_round_at_same_lsn() {
let before = DateTime::<Utc>::from(before);
let initdb_lsn = Lsn(0x10000);
let pitr_cutoff = Lsn(0x11000);
let disk_consistent_lsn = Lsn(initdb_lsn.0 * 2);
let logical_size = 0x42000;
let mut metrics = Vec::new();
let cache = HashMap::from([
@@ -123,10 +103,7 @@ fn startup_collected_timeline_metrics_nth_round_at_same_lsn() {
let snap = TimelineSnapshot {
loaded_at: (disk_consistent_lsn, init),
last_record_lsn: disk_consistent_lsn,
ancestor_lsn: Lsn(0),
current_exact_logical_size: Some(logical_size),
pitr_enabled: true,
pitr_cutoff: Some(pitr_cutoff),
current_exact_logical_size: Some(0x42000),
};
snap.to_metrics(tenant_id, timeline_id, now, &mut metrics, &cache);
@@ -136,18 +113,16 @@ fn startup_collected_timeline_metrics_nth_round_at_same_lsn() {
&[
MetricsKey::written_size_delta(tenant_id, timeline_id).from_until(just_before, now, 0),
MetricsKey::written_size(tenant_id, timeline_id).at(now, disk_consistent_lsn.0),
MetricsKey::written_size_since_parent(tenant_id, timeline_id)
.at(now, disk_consistent_lsn.0),
MetricsKey::pitr_history_size_since_parent(tenant_id, timeline_id)
.at(now, disk_consistent_lsn.0 - pitr_cutoff.0),
MetricsKey::timeline_logical_size(tenant_id, timeline_id).at(now, logical_size)
MetricsKey::timeline_logical_size(tenant_id, timeline_id).at(now, 0x42000)
]
);
}
/// Tests that written sizes do not regress across restarts.
#[test]
fn post_restart_written_sizes_with_rolled_back_last_record_lsn() {
// it can happen that we lose the inmemorylayer but have previously sent metrics and we
// should never go backwards
let tenant_id = TenantId::generate();
let timeline_id = TimelineId::generate();
@@ -165,10 +140,7 @@ fn post_restart_written_sizes_with_rolled_back_last_record_lsn() {
let snap = TimelineSnapshot {
loaded_at: (Lsn(50), at_restart),
last_record_lsn: Lsn(50),
ancestor_lsn: Lsn(0),
current_exact_logical_size: None,
pitr_enabled: true,
pitr_cutoff: Some(Lsn(20)),
};
let mut cache = HashMap::from([
@@ -197,8 +169,6 @@ fn post_restart_written_sizes_with_rolled_back_last_record_lsn() {
0
),
MetricsKey::written_size(tenant_id, timeline_id).at(now, 100),
MetricsKey::written_size_since_parent(tenant_id, timeline_id).at(now, 100),
MetricsKey::pitr_history_size_since_parent(tenant_id, timeline_id).at(now, 80),
]
);
@@ -213,157 +183,6 @@ fn post_restart_written_sizes_with_rolled_back_last_record_lsn() {
&[
MetricsKey::written_size_delta(tenant_id, timeline_id).from_until(now, later, 0),
MetricsKey::written_size(tenant_id, timeline_id).at(later, 100),
MetricsKey::written_size_since_parent(tenant_id, timeline_id).at(later, 100),
MetricsKey::pitr_history_size_since_parent(tenant_id, timeline_id).at(later, 80),
]
);
}
/// Tests that written sizes do not regress across restarts, even on child branches.
#[test]
fn post_restart_written_sizes_with_rolled_back_last_record_lsn_and_ancestor_lsn() {
let tenant_id = TenantId::generate();
let timeline_id = TimelineId::generate();
let [later, now, at_restart] = time_backwards();
// FIXME: tests would be so much easier if we did not need to juggle back and forth
// SystemTime and DateTime::<Utc> ... Could do the conversion only at upload time?
let now = DateTime::<Utc>::from(now);
let later = DateTime::<Utc>::from(later);
let before_restart = at_restart - std::time::Duration::from_secs(5 * 60);
let way_before = before_restart - std::time::Duration::from_secs(10 * 60);
let before_restart = DateTime::<Utc>::from(before_restart);
let way_before = DateTime::<Utc>::from(way_before);
let snap = TimelineSnapshot {
loaded_at: (Lsn(50), at_restart),
last_record_lsn: Lsn(50),
ancestor_lsn: Lsn(40),
current_exact_logical_size: None,
pitr_enabled: true,
pitr_cutoff: Some(Lsn(20)),
};
let mut cache = HashMap::from([
MetricsKey::written_size(tenant_id, timeline_id)
.at(before_restart, 100)
.to_kv_pair(),
MetricsKey::written_size_delta(tenant_id, timeline_id)
.from_until(
way_before,
before_restart,
// not taken into account, but the timestamps are important
999_999_999,
)
.to_kv_pair(),
]);
let mut metrics = Vec::new();
snap.to_metrics(tenant_id, timeline_id, now, &mut metrics, &cache);
assert_eq!(
metrics,
&[
MetricsKey::written_size_delta(tenant_id, timeline_id).from_until(
before_restart,
now,
0
),
MetricsKey::written_size(tenant_id, timeline_id).at(now, 100),
MetricsKey::written_size_since_parent(tenant_id, timeline_id).at(now, 60),
MetricsKey::pitr_history_size_since_parent(tenant_id, timeline_id).at(now, 60),
]
);
// now if we cache these metrics, and re-run while "still in recovery"
cache.extend(metrics.drain(..).map(|x| x.to_kv_pair()));
// "still in recovery", because our snapshot did not change
snap.to_metrics(tenant_id, timeline_id, later, &mut metrics, &cache);
assert_eq!(
metrics,
&[
MetricsKey::written_size_delta(tenant_id, timeline_id).from_until(now, later, 0),
MetricsKey::written_size(tenant_id, timeline_id).at(later, 100),
MetricsKey::written_size_since_parent(tenant_id, timeline_id).at(later, 60),
MetricsKey::pitr_history_size_since_parent(tenant_id, timeline_id).at(later, 60),
]
);
}
/// Tests that written sizes do not regress across restarts, even on child branches and
/// with a PITR cutoff after the branch point.
#[test]
fn post_restart_written_sizes_with_rolled_back_last_record_lsn_and_ancestor_lsn_and_pitr_cutoff() {
let tenant_id = TenantId::generate();
let timeline_id = TimelineId::generate();
let [later, now, at_restart] = time_backwards();
// FIXME: tests would be so much easier if we did not need to juggle back and forth
// SystemTime and DateTime::<Utc> ... Could do the conversion only at upload time?
let now = DateTime::<Utc>::from(now);
let later = DateTime::<Utc>::from(later);
let before_restart = at_restart - std::time::Duration::from_secs(5 * 60);
let way_before = before_restart - std::time::Duration::from_secs(10 * 60);
let before_restart = DateTime::<Utc>::from(before_restart);
let way_before = DateTime::<Utc>::from(way_before);
let snap = TimelineSnapshot {
loaded_at: (Lsn(50), at_restart),
last_record_lsn: Lsn(50),
ancestor_lsn: Lsn(30),
current_exact_logical_size: None,
pitr_enabled: true,
pitr_cutoff: Some(Lsn(40)),
};
let mut cache = HashMap::from([
MetricsKey::written_size(tenant_id, timeline_id)
.at(before_restart, 100)
.to_kv_pair(),
MetricsKey::written_size_delta(tenant_id, timeline_id)
.from_until(
way_before,
before_restart,
// not taken into account, but the timestamps are important
999_999_999,
)
.to_kv_pair(),
]);
let mut metrics = Vec::new();
snap.to_metrics(tenant_id, timeline_id, now, &mut metrics, &cache);
assert_eq!(
metrics,
&[
MetricsKey::written_size_delta(tenant_id, timeline_id).from_until(
before_restart,
now,
0
),
MetricsKey::written_size(tenant_id, timeline_id).at(now, 100),
MetricsKey::written_size_since_parent(tenant_id, timeline_id).at(now, 70),
MetricsKey::pitr_history_size_since_parent(tenant_id, timeline_id).at(now, 60),
]
);
// now if we cache these metrics, and re-run while "still in recovery"
cache.extend(metrics.drain(..).map(|x| x.to_kv_pair()));
// "still in recovery", because our snapshot did not change
snap.to_metrics(tenant_id, timeline_id, later, &mut metrics, &cache);
assert_eq!(
metrics,
&[
MetricsKey::written_size_delta(tenant_id, timeline_id).from_until(now, later, 0),
MetricsKey::written_size(tenant_id, timeline_id).at(later, 100),
MetricsKey::written_size_since_parent(tenant_id, timeline_id).at(later, 70),
MetricsKey::pitr_history_size_since_parent(tenant_id, timeline_id).at(later, 60),
]
);
}
@@ -382,10 +201,7 @@ fn post_restart_current_exact_logical_size_uses_cached() {
let snap = TimelineSnapshot {
loaded_at: (Lsn(50), at_restart),
last_record_lsn: Lsn(50),
ancestor_lsn: Lsn(0),
current_exact_logical_size: None,
pitr_enabled: true,
pitr_cutoff: None,
};
let cache = HashMap::from([MetricsKey::timeline_logical_size(tenant_id, timeline_id)
@@ -470,101 +286,16 @@ fn time_backwards<const N: usize>() -> [std::time::SystemTime; N] {
times
}
/// Tests that disabled PITR history does not yield any history size, even when the PITR cutoff
/// indicates otherwise.
#[test]
fn pitr_disabled_yields_no_history_size() {
let tenant_id = TenantId::generate();
let timeline_id = TimelineId::generate();
let mut metrics = Vec::new();
let cache = HashMap::new();
let initdb_lsn = Lsn(0x10000);
let pitr_cutoff = Lsn(0x11000);
let disk_consistent_lsn = Lsn(initdb_lsn.0 * 2);
let snap = TimelineSnapshot {
loaded_at: (disk_consistent_lsn, SystemTime::now()),
last_record_lsn: disk_consistent_lsn,
ancestor_lsn: Lsn(0),
current_exact_logical_size: None,
pitr_enabled: false,
pitr_cutoff: Some(pitr_cutoff),
};
let now = DateTime::<Utc>::from(SystemTime::now());
snap.to_metrics(tenant_id, timeline_id, now, &mut metrics, &cache);
assert_eq!(
metrics,
&[
MetricsKey::written_size_delta(tenant_id, timeline_id).from_until(
snap.loaded_at.1.into(),
now,
0
),
MetricsKey::written_size(tenant_id, timeline_id).at(now, disk_consistent_lsn.0),
MetricsKey::written_size_since_parent(tenant_id, timeline_id)
.at(now, disk_consistent_lsn.0),
MetricsKey::pitr_history_size_since_parent(tenant_id, timeline_id).at(now, 0),
]
);
}
/// Tests that uninitialized PITR cutoff does not emit any history size metric at all.
#[test]
fn pitr_uninitialized_does_not_emit_history_size() {
let tenant_id = TenantId::generate();
let timeline_id = TimelineId::generate();
let mut metrics = Vec::new();
let cache = HashMap::new();
let initdb_lsn = Lsn(0x10000);
let disk_consistent_lsn = Lsn(initdb_lsn.0 * 2);
let snap = TimelineSnapshot {
loaded_at: (disk_consistent_lsn, SystemTime::now()),
last_record_lsn: disk_consistent_lsn,
ancestor_lsn: Lsn(0),
current_exact_logical_size: None,
pitr_enabled: true,
pitr_cutoff: None,
};
let now = DateTime::<Utc>::from(SystemTime::now());
snap.to_metrics(tenant_id, timeline_id, now, &mut metrics, &cache);
assert_eq!(
metrics,
&[
MetricsKey::written_size_delta(tenant_id, timeline_id).from_until(
snap.loaded_at.1.into(),
now,
0
),
MetricsKey::written_size(tenant_id, timeline_id).at(now, disk_consistent_lsn.0),
MetricsKey::written_size_since_parent(tenant_id, timeline_id)
.at(now, disk_consistent_lsn.0),
]
);
}
pub(crate) const fn metric_examples_old(
tenant_id: TenantId,
timeline_id: TimelineId,
now: DateTime<Utc>,
before: DateTime<Utc>,
) -> [RawMetric; 7] {
) -> [RawMetric; 5] {
[
MetricsKey::written_size(tenant_id, timeline_id).at_old_format(now, 0),
MetricsKey::written_size_delta(tenant_id, timeline_id)
.from_until_old_format(before, now, 0),
MetricsKey::written_size_since_parent(tenant_id, timeline_id).at_old_format(now, 0),
MetricsKey::pitr_history_size_since_parent(tenant_id, timeline_id).at_old_format(now, 0),
MetricsKey::timeline_logical_size(tenant_id, timeline_id).at_old_format(now, 0),
MetricsKey::remote_storage_size(tenant_id).at_old_format(now, 0),
MetricsKey::synthetic_size(tenant_id).at_old_format(now, 1),
@@ -576,12 +307,10 @@ pub(crate) const fn metric_examples(
timeline_id: TimelineId,
now: DateTime<Utc>,
before: DateTime<Utc>,
) -> [NewRawMetric; 7] {
) -> [NewRawMetric; 5] {
[
MetricsKey::written_size(tenant_id, timeline_id).at(now, 0),
MetricsKey::written_size_delta(tenant_id, timeline_id).from_until(before, now, 0),
MetricsKey::written_size_since_parent(tenant_id, timeline_id).at(now, 0),
MetricsKey::pitr_history_size_since_parent(tenant_id, timeline_id).at(now, 0),
MetricsKey::timeline_logical_size(tenant_id, timeline_id).at(now, 0),
MetricsKey::remote_storage_size(tenant_id).at(now, 0),
MetricsKey::synthetic_size(tenant_id).at(now, 1),

View File

@@ -513,14 +513,6 @@ mod tests {
line!(),
r#"{"type":"incremental","start_time":"2023-09-14T00:00:00.123456789Z","stop_time":"2023-09-15T00:00:00.123456789Z","metric":"written_data_bytes_delta","idempotency_key":"2023-09-15 00:00:00.123456789 UTC-1-0000","value":0,"tenant_id":"00000000000000000000000000000000","timeline_id":"ffffffffffffffffffffffffffffffff"}"#,
),
(
line!(),
r#"{"type":"absolute","time":"2023-09-15T00:00:00.123456789Z","metric":"written_size_since_parent","idempotency_key":"2023-09-15 00:00:00.123456789 UTC-1-0000","value":0,"tenant_id":"00000000000000000000000000000000","timeline_id":"ffffffffffffffffffffffffffffffff"}"#,
),
(
line!(),
r#"{"type":"absolute","time":"2023-09-15T00:00:00.123456789Z","metric":"pitr_history_size_since_parent","idempotency_key":"2023-09-15 00:00:00.123456789 UTC-1-0000","value":0,"tenant_id":"00000000000000000000000000000000","timeline_id":"ffffffffffffffffffffffffffffffff"}"#,
),
(
line!(),
r#"{"type":"absolute","time":"2023-09-15T00:00:00.123456789Z","metric":"timeline_logical_size","idempotency_key":"2023-09-15 00:00:00.123456789 UTC-1-0000","value":0,"tenant_id":"00000000000000000000000000000000","timeline_id":"ffffffffffffffffffffffffffffffff"}"#,
@@ -568,7 +560,7 @@ mod tests {
assert_eq!(upgraded_samples, new_samples);
}
fn metric_samples_old() -> [RawMetric; 7] {
fn metric_samples_old() -> [RawMetric; 5] {
let tenant_id = TenantId::from_array([0; 16]);
let timeline_id = TimelineId::from_array([0xff; 16]);
@@ -580,7 +572,7 @@ mod tests {
super::super::metrics::metric_examples_old(tenant_id, timeline_id, now, before)
}
fn metric_samples() -> [NewRawMetric; 7] {
fn metric_samples() -> [NewRawMetric; 5] {
let tenant_id = TenantId::from_array([0; 16]);
let timeline_id = TimelineId::from_array([0xff; 16]);

View File

@@ -7,7 +7,7 @@ use pageserver_api::models::ShardImportStatus;
use pageserver_api::shard::TenantShardId;
use pageserver_api::upcall_api::{
PutTimelineImportStatusRequest, ReAttachRequest, ReAttachResponse, ReAttachResponseTenant,
TimelineImportStatusRequest, ValidateRequest, ValidateRequestTenant, ValidateResponse,
ValidateRequest, ValidateRequestTenant, ValidateResponse,
};
use reqwest::Certificate;
use serde::Serialize;
@@ -51,15 +51,13 @@ pub trait StorageControllerUpcallApi {
&self,
tenant_shard_id: TenantShardId,
timeline_id: TimelineId,
generation: Generation,
status: ShardImportStatus,
) -> impl Future<Output = Result<(), RetryForeverError>> + Send;
fn get_timeline_import_status(
&self,
tenant_shard_id: TenantShardId,
timeline_id: TimelineId,
generation: Generation,
) -> impl Future<Output = Result<ShardImportStatus, RetryForeverError>> + Send;
) -> impl Future<Output = Result<Option<ShardImportStatus>, RetryForeverError>> + Send;
}
impl StorageControllerUpcallClient {
@@ -104,7 +102,6 @@ impl StorageControllerUpcallClient {
&self,
url: &url::Url,
request: R,
method: reqwest::Method,
) -> Result<T, RetryForeverError>
where
R: Serialize,
@@ -114,7 +111,7 @@ impl StorageControllerUpcallClient {
|| async {
let response = self
.http_client
.request(method.clone(), url.clone())
.post(url.clone())
.json(&request)
.send()
.await?;
@@ -223,9 +220,7 @@ impl StorageControllerUpcallApi for StorageControllerUpcallClient {
register: register.clone(),
};
let response: ReAttachResponse = self
.retry_http_forever(&url, request, reqwest::Method::POST)
.await?;
let response: ReAttachResponse = self.retry_http_forever(&url, request).await?;
tracing::info!(
"Received re-attach response with {} tenants (node {}, register: {:?})",
response.tenants.len(),
@@ -278,9 +273,7 @@ impl StorageControllerUpcallApi for StorageControllerUpcallClient {
return Err(RetryForeverError::ShuttingDown);
}
let response: ValidateResponse = self
.retry_http_forever(&url, request, reqwest::Method::POST)
.await?;
let response: ValidateResponse = self.retry_http_forever(&url, request).await?;
for rt in response.tenants {
result.insert(rt.id, rt.valid);
}
@@ -299,7 +292,6 @@ impl StorageControllerUpcallApi for StorageControllerUpcallClient {
&self,
tenant_shard_id: TenantShardId,
timeline_id: TimelineId,
generation: Generation,
status: ShardImportStatus,
) -> Result<(), RetryForeverError> {
let url = self
@@ -310,12 +302,10 @@ impl StorageControllerUpcallApi for StorageControllerUpcallClient {
let request = PutTimelineImportStatusRequest {
tenant_shard_id,
timeline_id,
generation,
status,
};
self.retry_http_forever(&url, request, reqwest::Method::POST)
.await
self.retry_http_forever(&url, request).await
}
#[tracing::instrument(skip_all)] // so that warning logs from retry_http_forever have context
@@ -323,22 +313,33 @@ impl StorageControllerUpcallApi for StorageControllerUpcallClient {
&self,
tenant_shard_id: TenantShardId,
timeline_id: TimelineId,
generation: Generation,
) -> Result<ShardImportStatus, RetryForeverError> {
) -> Result<Option<ShardImportStatus>, RetryForeverError> {
let url = self
.base_url
.join("timeline_import_status")
.join(format!("timeline_import_status/{}/{}", tenant_shard_id, timeline_id).as_str())
.expect("Failed to build path");
let request = TimelineImportStatusRequest {
tenant_shard_id,
timeline_id,
generation,
};
Ok(backoff::retry(
|| async {
let response = self.http_client.get(url.clone()).send().await?;
let response: ShardImportStatus = self
.retry_http_forever(&url, request, reqwest::Method::GET)
.await?;
Ok(response)
if let Err(err) = response.error_for_status_ref() {
if matches!(err.status(), Some(reqwest::StatusCode::NOT_FOUND)) {
return Ok(None);
} else {
return Err(err);
}
}
response.json::<ShardImportStatus>().await.map(Some)
},
|_| false,
3,
u32::MAX,
"storage controller upcall",
&self.cancel,
)
.await
.ok_or(RetryForeverError::ShuttingDown)?
.expect("We retry forever, this should never be reached"))
}
}

View File

@@ -793,7 +793,6 @@ mod test {
&self,
_tenant_shard_id: TenantShardId,
_timeline_id: TimelineId,
_generation: Generation,
_status: pageserver_api::models::ShardImportStatus,
) -> Result<(), RetryForeverError> {
unimplemented!()
@@ -803,8 +802,7 @@ mod test {
&self,
_tenant_shard_id: TenantShardId,
_timeline_id: TimelineId,
_generation: Generation,
) -> Result<ShardImportStatus, RetryForeverError> {
) -> Result<Option<ShardImportStatus>, RetryForeverError> {
unimplemented!()
}
}

View File

@@ -837,30 +837,7 @@ async fn collect_eviction_candidates(
continue;
}
let info = tl.get_local_layers_for_disk_usage_eviction().await;
debug!(
tenant_id=%tl.tenant_shard_id.tenant_id,
shard_id=%tl.tenant_shard_id.shard_slug(),
timeline_id=%tl.timeline_id,
"timeline resident layers count: {}", info.resident_layers.len()
);
tenant_candidates.extend(info.resident_layers.into_iter());
max_layer_size = max_layer_size.max(info.max_layer_size.unwrap_or(0));
if cancel.is_cancelled() {
return Ok(EvictionCandidates::Cancelled);
}
}
// Also consider layers of timelines being imported for eviction
for tl in tenant.list_importing_timelines() {
let info = tl.timeline.get_local_layers_for_disk_usage_eviction().await;
debug!(
tenant_id=%tl.timeline.tenant_shard_id.tenant_id,
shard_id=%tl.timeline.tenant_shard_id.shard_slug(),
timeline_id=%tl.timeline.timeline_id,
"timeline resident layers count: {}", info.resident_layers.len()
);
debug!(tenant_id=%tl.tenant_shard_id.tenant_id, shard_id=%tl.tenant_shard_id.shard_slug(), timeline_id=%tl.timeline_id, "timeline resident layers count: {}", info.resident_layers.len());
tenant_candidates.extend(info.resident_layers.into_iter());
max_layer_size = max_layer_size.max(info.max_layer_size.unwrap_or(0));

View File

@@ -1,104 +0,0 @@
use std::{collections::HashMap, sync::Arc, time::Duration};
use posthog_client_lite::{
FeatureResolverBackgroundLoop, PostHogClientConfig, PostHogEvaluationError,
};
use tokio_util::sync::CancellationToken;
use utils::id::TenantId;
use crate::config::PageServerConf;
#[derive(Clone)]
pub struct FeatureResolver {
inner: Option<Arc<FeatureResolverBackgroundLoop>>,
}
impl FeatureResolver {
pub fn new_disabled() -> Self {
Self { inner: None }
}
pub fn spawn(
conf: &PageServerConf,
shutdown_pageserver: CancellationToken,
handle: &tokio::runtime::Handle,
) -> anyhow::Result<Self> {
// DO NOT block in this function: make it return as fast as possible to avoid startup delays.
if let Some(posthog_config) = &conf.posthog_config {
let inner = FeatureResolverBackgroundLoop::new(
PostHogClientConfig {
server_api_key: posthog_config.server_api_key.clone(),
client_api_key: posthog_config.client_api_key.clone(),
project_id: posthog_config.project_id.clone(),
private_api_url: posthog_config.private_api_url.clone(),
public_api_url: posthog_config.public_api_url.clone(),
},
shutdown_pageserver,
);
let inner = Arc::new(inner);
// TODO: make this configurable
inner.clone().spawn(handle, Duration::from_secs(60));
Ok(FeatureResolver { inner: Some(inner) })
} else {
Ok(FeatureResolver { inner: None })
}
}
/// Evaluate a multivariate feature flag. Currently, we do not support any properties.
///
/// Error handling: the caller should inspect the error and decide the behavior when a feature flag
/// cannot be evaluated (i.e., default to false if it cannot be resolved). The error should *not* be
/// propagated beyond where the feature flag gets resolved.
pub fn evaluate_multivariate(
&self,
flag_key: &str,
tenant_id: TenantId,
) -> Result<String, PostHogEvaluationError> {
if let Some(inner) = &self.inner {
inner.feature_store().evaluate_multivariate(
flag_key,
&tenant_id.to_string(),
&HashMap::new(),
)
} else {
Err(PostHogEvaluationError::NotAvailable(
"PostHog integration is not enabled".to_string(),
))
}
}
/// Evaluate a boolean feature flag. Currently, we do not support any properties.
///
/// Returns `Ok(())` if the flag is evaluated to true, otherwise returns an error.
///
/// Error handling: the caller should inspect the error and decide the behavior when a feature flag
/// cannot be evaluated (i.e., default to false if it cannot be resolved). The error should *not* be
/// propagated beyond where the feature flag gets resolved.
pub fn evaluate_boolean(
&self,
flag_key: &str,
tenant_id: TenantId,
) -> Result<(), PostHogEvaluationError> {
if let Some(inner) = &self.inner {
inner.feature_store().evaluate_boolean(
flag_key,
&tenant_id.to_string(),
&HashMap::new(),
)
} else {
Err(PostHogEvaluationError::NotAvailable(
"PostHog integration is not enabled".to_string(),
))
}
}
pub fn is_feature_flag_boolean(&self, flag_key: &str) -> Result<bool, PostHogEvaluationError> {
if let Some(inner) = &self.inner {
inner.feature_store().is_feature_flag_boolean(flag_key)
} else {
Err(PostHogEvaluationError::NotAvailable(
"PostHog integration is not enabled".to_string(),
))
}
}
}

View File

@@ -353,33 +353,6 @@ paths:
"200":
description: OK
/v1/tenant/{tenant_shard_id}/timeline/{timeline_id}/mark_invisible:
parameters:
- name: tenant_shard_id
in: path
required: true
schema:
type: string
- name: timeline_id
in: path
required: true
schema:
type: string
format: hex
put:
requestBody:
content:
application/json:
schema:
type: object
properties:
is_visible:
type: boolean
default: false
responses:
"200":
description: OK
/v1/tenant/{tenant_shard_id}/location_config:
parameters:
- name: tenant_shard_id
@@ -653,8 +626,6 @@ paths:
format: hex
pg_version:
type: integer
read_only:
type: boolean
existing_initdb_timeline_id:
type: string
format: hex

View File

@@ -370,18 +370,6 @@ impl From<crate::tenant::secondary::SecondaryTenantError> for ApiError {
}
}
impl From<crate::tenant::FinalizeTimelineImportError> for ApiError {
fn from(err: crate::tenant::FinalizeTimelineImportError) -> ApiError {
use crate::tenant::FinalizeTimelineImportError::*;
match err {
ImportTaskStillRunning => {
ApiError::ResourceUnavailable("Import task still running".into())
}
ShuttingDown => ApiError::ShuttingDown,
}
}
}
// Helper function to construct a TimelineInfo struct for a timeline
async fn build_timeline_info(
timeline: &Arc<Timeline>,
@@ -461,7 +449,7 @@ async fn build_timeline_info_common(
// Internally we distinguish between the planned GC cutoff (PITR point) and the "applied" GC cutoff (where we
// actually trimmed data to), which can pass each other when PITR is changed.
let min_readable_lsn = std::cmp::max(
timeline.get_gc_cutoff_lsn().unwrap_or_default(),
timeline.get_gc_cutoff_lsn(),
*timeline.get_applied_gc_cutoff_lsn(),
);
@@ -584,7 +572,6 @@ async fn timeline_create_handler(
TimelineCreateRequestMode::Branch {
ancestor_timeline_id,
ancestor_start_lsn,
read_only: _,
pg_version: _,
} => tenant::CreateTimelineParams::Branch(tenant::CreateTimelineParamsBranch {
new_timeline_id,
@@ -3212,7 +3199,7 @@ async fn list_aux_files(
.await?;
let io_concurrency = IoConcurrency::spawn_from_conf(
state.conf.get_vectored_concurrent_io,
state.conf,
timeline.gate.enter().map_err(|_| ApiError::Cancelled)?,
);
@@ -3545,7 +3532,10 @@ async fn activate_post_import_handler(
tenant.wait_to_become_active(ACTIVE_TENANT_TIMEOUT).await?;
tenant.finalize_importing_timeline(timeline_id).await?;
tenant
.finalize_importing_timeline(timeline_id)
.await
.map_err(ApiError::InternalServerError)?;
match tenant.get_timeline(timeline_id, false) {
Ok(_timeline) => {
@@ -3663,46 +3653,6 @@ async fn read_tar_eof(mut reader: (impl tokio::io::AsyncRead + Unpin)) -> anyhow
Ok(())
}
async fn tenant_evaluate_feature_flag(
request: Request<Body>,
_cancel: CancellationToken,
) -> Result<Response<Body>, ApiError> {
let tenant_shard_id: TenantShardId = parse_request_param(&request, "tenant_shard_id")?;
check_permission(&request, Some(tenant_shard_id.tenant_id))?;
let flag: String = must_parse_query_param(&request, "flag")?;
let as_type: String = must_parse_query_param(&request, "as")?;
let state = get_state(&request);
async {
let tenant = state
.tenant_manager
.get_attached_tenant_shard(tenant_shard_id)?;
if as_type == "boolean" {
let result = tenant.feature_resolver.evaluate_boolean(&flag, tenant_shard_id.tenant_id);
let result = result.map(|_| true).map_err(|e| e.to_string());
json_response(StatusCode::OK, result)
} else if as_type == "multivariate" {
let result = tenant.feature_resolver.evaluate_multivariate(&flag, tenant_shard_id.tenant_id).map_err(|e| e.to_string());
json_response(StatusCode::OK, result)
} else {
// Auto infer the type of the feature flag.
let is_boolean = tenant.feature_resolver.is_feature_flag_boolean(&flag).map_err(|e| ApiError::InternalServerError(anyhow::anyhow!("{e}")))?;
if is_boolean {
let result = tenant.feature_resolver.evaluate_boolean(&flag, tenant_shard_id.tenant_id);
let result = result.map(|_| true).map_err(|e| e.to_string());
json_response(StatusCode::OK, result)
} else {
let result = tenant.feature_resolver.evaluate_multivariate(&flag, tenant_shard_id.tenant_id).map_err(|e| e.to_string());
json_response(StatusCode::OK, result)
}
}
}
.instrument(info_span!("tenant_evaluate_feature_flag", tenant_id = %tenant_shard_id.tenant_id, shard_id = %tenant_shard_id.shard_slug()))
.await
}
/// Common functionality of all the HTTP API handlers.
///
/// - Adds a tracing span to each request (by `request_span`)
@@ -4079,8 +4029,5 @@ pub fn make_router(
"/v1/tenant/:tenant_shard_id/timeline/:timeline_id/activate_post_import",
|r| api_handler(r, activate_post_import_handler),
)
.get("/v1/tenant/:tenant_shard_id/feature_flag", |r| {
api_handler(r, tenant_evaluate_feature_flag)
})
.any(handler_404))
}

View File

@@ -3,14 +3,12 @@
mod auth;
pub mod basebackup;
pub mod basebackup_cache;
pub mod config;
pub mod consumption_metrics;
pub mod context;
pub mod controller_upcall_client;
pub mod deletion_queue;
pub mod disk_usage_eviction_task;
pub mod feature_resolver;
pub mod http;
pub mod import_datadir;
pub mod l0_flush;
@@ -85,7 +83,6 @@ pub async fn shutdown_pageserver(
http_listener: HttpEndpointListener,
https_listener: Option<HttpsEndpointListener>,
page_service: page_service::Listener,
grpc_task: Option<CancellableTask>,
consumption_metrics_worker: ConsumptionMetricsTasks,
disk_usage_eviction_task: Option<DiskUsageEvictionTask>,
tenant_manager: &TenantManager,
@@ -179,16 +176,6 @@ pub async fn shutdown_pageserver(
)
.await;
// Shut down the gRPC server task, including request handlers.
if let Some(grpc_task) = grpc_task {
timed(
grpc_task.shutdown(),
"shutdown gRPC PageRequestHandler",
Duration::from_secs(3),
)
.await;
}
// Shut down all the tenants. This flushes everything to disk and kills
// the checkpoint and GC tasks.
timed(

View File

@@ -843,50 +843,23 @@ pub(crate) static COMPRESSION_IMAGE_OUTPUT_BYTES: Lazy<IntCounter> = Lazy::new(|
.expect("failed to define a metric")
});
pub(crate) static RELSIZE_LATEST_CACHE_ENTRIES: Lazy<UIntGauge> = Lazy::new(|| {
pub(crate) static RELSIZE_CACHE_ENTRIES: Lazy<UIntGauge> = Lazy::new(|| {
register_uint_gauge!(
"pageserver_relsize_latest_cache_entries",
"Number of entries in the latest relation size cache",
"pageserver_relsize_cache_entries",
"Number of entries in the relation size cache",
)
.expect("failed to define a metric")
});
pub(crate) static RELSIZE_LATEST_CACHE_HITS: Lazy<IntCounter> = Lazy::new(|| {
pub(crate) static RELSIZE_CACHE_HITS: Lazy<IntCounter> = Lazy::new(|| {
register_int_counter!("pageserver_relsize_cache_hits", "Relation size cache hits",)
.expect("failed to define a metric")
});
pub(crate) static RELSIZE_CACHE_MISSES: Lazy<IntCounter> = Lazy::new(|| {
register_int_counter!(
"pageserver_relsize_latest_cache_hits",
"Latest relation size cache hits",
)
.expect("failed to define a metric")
});
pub(crate) static RELSIZE_LATEST_CACHE_MISSES: Lazy<IntCounter> = Lazy::new(|| {
register_int_counter!(
"pageserver_relsize_latest_cache_misses",
"Relation size latest cache misses",
)
.expect("failed to define a metric")
});
pub(crate) static RELSIZE_SNAPSHOT_CACHE_ENTRIES: Lazy<UIntGauge> = Lazy::new(|| {
register_uint_gauge!(
"pageserver_relsize_snapshot_cache_entries",
"Number of entries in the pitr relation size cache",
)
.expect("failed to define a metric")
});
pub(crate) static RELSIZE_SNAPSHOT_CACHE_HITS: Lazy<IntCounter> = Lazy::new(|| {
register_int_counter!(
"pageserver_relsize_snapshot_cache_hits",
"Pitr relation size cache hits",
)
.expect("failed to define a metric")
});
pub(crate) static RELSIZE_SNAPSHOT_CACHE_MISSES: Lazy<IntCounter> = Lazy::new(|| {
register_int_counter!(
"pageserver_relsize_snapshot_cache_misses",
"Relation size snapshot cache misses",
"pageserver_relsize_cache_misses",
"Relation size cache misses",
)
.expect("failed to define a metric")
});
@@ -1066,15 +1039,6 @@ pub(crate) static TENANT_SYNTHETIC_SIZE_METRIC: Lazy<UIntGaugeVec> = Lazy::new(|
.expect("Failed to register pageserver_tenant_synthetic_cached_size_bytes metric")
});
pub(crate) static TENANT_OFFLOADED_TIMELINES: Lazy<UIntGaugeVec> = Lazy::new(|| {
register_uint_gauge_vec!(
"pageserver_tenant_offloaded_timelines",
"Number of offloaded timelines of a tenant",
&["tenant_id", "shard_id"]
)
.expect("Failed to register pageserver_tenant_offloaded_timelines metric")
});
pub(crate) static EVICTION_ITERATION_DURATION: Lazy<HistogramVec> = Lazy::new(|| {
register_histogram_vec!(
"pageserver_eviction_iteration_duration_seconds_global",
@@ -1312,44 +1276,11 @@ impl EvictionsWithLowResidenceDuration {
//
// Roughly logarithmic scale.
const STORAGE_IO_TIME_BUCKETS: &[f64] = &[
0.00005, // 50us
0.00006, // 60us
0.00007, // 70us
0.00008, // 80us
0.00009, // 90us
0.0001, // 100us
0.000110, // 110us
0.000120, // 120us
0.000130, // 130us
0.000140, // 140us
0.000150, // 150us
0.000160, // 160us
0.000170, // 170us
0.000180, // 180us
0.000190, // 190us
0.000200, // 200us
0.000210, // 210us
0.000220, // 220us
0.000230, // 230us
0.000240, // 240us
0.000250, // 250us
0.000300, // 300us
0.000350, // 350us
0.000400, // 400us
0.000450, // 450us
0.000500, // 500us
0.000600, // 600us
0.000700, // 700us
0.000800, // 800us
0.000900, // 900us
0.001000, // 1ms
0.002000, // 2ms
0.003000, // 3ms
0.004000, // 4ms
0.005000, // 5ms
0.01000, // 10ms
0.02000, // 20ms
0.05000, // 50ms
0.000030, // 30 usec
0.001000, // 1000 usec
0.030, // 30 ms
1.000, // 1000 ms
30.000, // 30000 ms
];
/// VirtualFile fs operation variants.
@@ -2267,10 +2198,8 @@ impl BasebackupQueryTimeOngoingRecording<'_> {
// If you want to change categorize of a specific error, also change it in `log_query_error`.
let metric = match res {
Ok(_) => &self.parent.ok,
Err(QueryError::Shutdown) | Err(QueryError::Reconnect) => {
// Do not observe ok/err for shutdown/reconnect.
// Reconnect error might be raised when the operation is waiting for LSN and the tenant shutdown interrupts
// the operation. A reconnect error will be issued and the client will retry.
Err(QueryError::Shutdown) => {
// Do not observe ok/err for shutdown
return;
}
Err(QueryError::Disconnected(ConnectionError::Io(io_error)))
@@ -3595,14 +3524,11 @@ impl TimelineMetrics {
}
pub(crate) fn remove_tenant_metrics(tenant_shard_id: &TenantShardId) {
let tid = tenant_shard_id.tenant_id.to_string();
let shard_id = tenant_shard_id.shard_slug().to_string();
// Only shard zero deals in synthetic sizes
if tenant_shard_id.is_shard_zero() {
let tid = tenant_shard_id.tenant_id.to_string();
let _ = TENANT_SYNTHETIC_SIZE_METRIC.remove_label_values(&[&tid]);
}
let _ = TENANT_OFFLOADED_TIMELINES.remove_label_values(&[&tid, &shard_id]);
tenant_throttling::remove_tenant_metrics(tenant_shard_id);
@@ -4394,42 +4320,6 @@ pub(crate) fn set_tokio_runtime_setup(setup: &str, num_threads: NonZeroUsize) {
.set(u64::try_from(num_threads.get()).unwrap());
}
pub(crate) static BASEBACKUP_CACHE_READ: Lazy<IntCounterVec> = Lazy::new(|| {
register_int_counter_vec!(
"pageserver_basebackup_cache_read_total",
"Number of read accesses to the basebackup cache grouped by hit/miss/error",
&["result"]
)
.expect("failed to define a metric")
});
pub(crate) static BASEBACKUP_CACHE_PREPARE: Lazy<IntCounterVec> = Lazy::new(|| {
register_int_counter_vec!(
"pageserver_basebackup_cache_prepare_total",
"Number of prepare requests processed by the basebackup cache grouped by ok/skip/error",
&["result"]
)
.expect("failed to define a metric")
});
pub(crate) static BASEBACKUP_CACHE_ENTRIES: Lazy<IntGauge> = Lazy::new(|| {
register_int_gauge!(
"pageserver_basebackup_cache_entries_total",
"Number of entries in the basebackup cache"
)
.expect("failed to define a metric")
});
// FIXME: Support basebackup cache size metrics.
#[allow(dead_code)]
pub(crate) static BASEBACKUP_CACHE_SIZE: Lazy<IntGauge> = Lazy::new(|| {
register_int_gauge!(
"pageserver_basebackup_cache_size_bytes",
"Total size of all basebackup cache entries on disk in bytes"
)
.expect("failed to define a metric")
});
static PAGESERVER_CONFIG_IGNORED_ITEMS: Lazy<UIntGaugeVec> = Lazy::new(|| {
register_uint_gauge_vec!(
"pageserver_config_ignored_items",

View File

@@ -4,21 +4,21 @@
use std::borrow::Cow;
use std::num::NonZeroUsize;
use std::os::fd::AsRawFd;
use std::pin::Pin;
use std::str::FromStr;
use std::sync::Arc;
use std::time::{Duration, Instant, SystemTime};
use std::{io, str};
use crate::PERF_TRACE_TARGET;
use anyhow::{Context, bail};
use async_compression::tokio::write::GzipEncoder;
use bytes::Buf;
use futures::{FutureExt, Stream};
use futures::FutureExt;
use itertools::Itertools;
use jsonwebtoken::TokenData;
use once_cell::sync::OnceCell;
use pageserver_api::config::{
GetVectoredConcurrentIo, PageServicePipeliningConfig, PageServicePipeliningConfigPipelined,
PageServicePipeliningConfig, PageServicePipeliningConfigPipelined,
PageServiceProtocolPipelinedBatchingStrategy, PageServiceProtocolPipelinedExecutionStrategy,
};
use pageserver_api::key::rel_block_to_key;
@@ -31,7 +31,6 @@ use pageserver_api::models::{
};
use pageserver_api::reltag::SlruKind;
use pageserver_api::shard::TenantShardId;
use pageserver_page_api::proto;
use postgres_backend::{
AuthType, PostgresBackend, PostgresBackendReader, QueryError, is_expected_io_error,
};
@@ -43,21 +42,18 @@ use strum_macros::IntoStaticStr;
use tokio::io::{AsyncRead, AsyncWrite, AsyncWriteExt, BufWriter};
use tokio::task::JoinHandle;
use tokio_util::sync::CancellationToken;
use tonic::service::Interceptor as _;
use tracing::*;
use utils::auth::{Claims, Scope, SwappableJwtAuth};
use utils::failpoint_support;
use utils::id::{TenantId, TenantTimelineId, TimelineId};
use utils::id::{TenantId, TimelineId};
use utils::logging::log_slow;
use utils::lsn::Lsn;
use utils::shard::ShardIndex;
use utils::simple_rcu::RcuReadGuard;
use utils::sync::gate::{Gate, GateGuard};
use utils::sync::spsc_fold;
use crate::auth::check_permission;
use crate::basebackup::{self, BasebackupError};
use crate::basebackup_cache::BasebackupCache;
use crate::basebackup::BasebackupError;
use crate::config::PageServerConf;
use crate::context::{
DownloadBehavior, PerfInstrumentFutureExt, RequestContext, RequestContextBuilder,
@@ -66,7 +62,7 @@ use crate::metrics::{
self, COMPUTE_COMMANDS_COUNTERS, ComputeCommandKind, GetPageBatchBreakReason, LIVE_CONNECTIONS,
SmgrOpTimer, TimelineMetrics,
};
use crate::pgdatadir_mapping::{LsnRange, Version};
use crate::pgdatadir_mapping::Version;
use crate::span::{
debug_assert_current_span_has_tenant_and_timeline_id,
debug_assert_current_span_has_tenant_and_timeline_id_no_shard_id,
@@ -78,7 +74,7 @@ use crate::tenant::mgr::{
use crate::tenant::storage_layer::IoConcurrency;
use crate::tenant::timeline::{self, WaitLsnError};
use crate::tenant::{GetTimelineError, PageReconstructError, Timeline};
use crate::{CancellableTask, PERF_TRACE_TARGET, timed_after_cancellation};
use crate::{basebackup, timed_after_cancellation};
/// How long we may wait for a [`crate::tenant::mgr::TenantSlot::InProgress`]` and/or a [`crate::tenant::TenantShard`] which
/// is not yet in state [`TenantState::Active`].
@@ -89,26 +85,6 @@ const ACTIVE_TENANT_TIMEOUT: Duration = Duration::from_millis(30000);
/// Threshold at which to log slow GetPage requests.
const LOG_SLOW_GETPAGE_THRESHOLD: Duration = Duration::from_secs(30);
/// The idle time before sending TCP keepalive probes for gRPC connections. The
/// interval and timeout between each probe is configured via sysctl. This
/// allows detecting dead connections sooner.
const GRPC_TCP_KEEPALIVE_TIME: Duration = Duration::from_secs(60);
/// Whether to enable TCP nodelay for gRPC connections. This disables Nagle's
/// algorithm, which can cause latency spikes for small messages.
const GRPC_TCP_NODELAY: bool = true;
/// The interval between HTTP2 keepalive pings. This allows shutting down server
/// tasks when clients are unresponsive.
const GRPC_HTTP2_KEEPALIVE_INTERVAL: Duration = Duration::from_secs(30);
/// The timeout for HTTP2 keepalive pings. Should be <= GRPC_KEEPALIVE_INTERVAL.
const GRPC_HTTP2_KEEPALIVE_TIMEOUT: Duration = Duration::from_secs(20);
/// Number of concurrent gRPC streams per TCP connection. We expect something
/// like 8 GetPage streams per connections, plus any unary requests.
const GRPC_MAX_CONCURRENT_STREAMS: u32 = 256;
///////////////////////////////////////////////////////////////////////////////
pub struct Listener {
@@ -131,7 +107,6 @@ pub fn spawn(
perf_trace_dispatch: Option<Dispatch>,
tcp_listener: tokio::net::TcpListener,
tls_config: Option<Arc<rustls::ServerConfig>>,
basebackup_cache: Arc<BasebackupCache>,
) -> Listener {
let cancel = CancellationToken::new();
let libpq_ctx = RequestContext::todo_child(
@@ -153,7 +128,6 @@ pub fn spawn(
conf.pg_auth_type,
tls_config,
conf.page_service_pipelining.clone(),
basebackup_cache,
libpq_ctx,
cancel.clone(),
)
@@ -163,94 +137,6 @@ pub fn spawn(
Listener { cancel, task }
}
/// Spawns a gRPC server for the page service.
///
/// TODO: this doesn't support TLS. We need TLS reloading via ReloadingCertificateResolver, so we
/// need to reimplement the TCP+TLS accept loop ourselves.
pub fn spawn_grpc(
conf: &'static PageServerConf,
tenant_manager: Arc<TenantManager>,
auth: Option<Arc<SwappableJwtAuth>>,
perf_trace_dispatch: Option<Dispatch>,
listener: std::net::TcpListener,
basebackup_cache: Arc<BasebackupCache>,
) -> anyhow::Result<CancellableTask> {
let cancel = CancellationToken::new();
let ctx = RequestContextBuilder::new(TaskKind::PageRequestHandler)
.download_behavior(DownloadBehavior::Download)
.perf_span_dispatch(perf_trace_dispatch)
.detached_child();
let gate = Gate::default();
// Set up the TCP socket. We take a preconfigured TcpListener to bind the
// port early during startup.
let incoming = {
let _runtime = COMPUTE_REQUEST_RUNTIME.enter(); // required by TcpListener::from_std
listener.set_nonblocking(true)?;
tonic::transport::server::TcpIncoming::from(tokio::net::TcpListener::from_std(listener)?)
.with_nodelay(Some(GRPC_TCP_NODELAY))
.with_keepalive(Some(GRPC_TCP_KEEPALIVE_TIME))
};
// Set up the gRPC server.
//
// TODO: consider tuning window sizes.
// TODO: wire up tracing.
let mut server = tonic::transport::Server::builder()
.http2_keepalive_interval(Some(GRPC_HTTP2_KEEPALIVE_INTERVAL))
.http2_keepalive_timeout(Some(GRPC_HTTP2_KEEPALIVE_TIMEOUT))
.max_concurrent_streams(Some(GRPC_MAX_CONCURRENT_STREAMS));
// Main page service.
let page_service_handler = PageServerHandler::new(
tenant_manager,
auth.clone(),
PageServicePipeliningConfig::Serial, // TODO: unused with gRPC
conf.get_vectored_concurrent_io,
ConnectionPerfSpanFields::default(),
basebackup_cache,
ctx,
cancel.clone(),
gate.enter().expect("just created"),
);
let mut tenant_interceptor = TenantMetadataInterceptor;
let mut auth_interceptor = TenantAuthInterceptor::new(auth);
let interceptors = move |mut req: tonic::Request<()>| {
req = tenant_interceptor.call(req)?;
req = auth_interceptor.call(req)?;
Ok(req)
};
let page_service =
proto::PageServiceServer::with_interceptor(page_service_handler, interceptors);
let server = server.add_service(page_service);
// Reflection service for use with e.g. grpcurl.
let reflection_service = tonic_reflection::server::Builder::configure()
.register_encoded_file_descriptor_set(proto::FILE_DESCRIPTOR_SET)
.build_v1()?;
let server = server.add_service(reflection_service);
// Spawn server task.
let task_cancel = cancel.clone();
let task = COMPUTE_REQUEST_RUNTIME.spawn(task_mgr::exit_on_panic_or_error(
"grpc listener",
async move {
let result = server
.serve_with_incoming_shutdown(incoming, task_cancel.cancelled())
.await;
if result.is_ok() {
// TODO: revisit shutdown logic once page service is implemented.
gate.close().await;
}
result
},
));
Ok(CancellableTask { task, cancel })
}
impl Listener {
pub async fn stop_accepting(self) -> Connections {
self.cancel.cancel();
@@ -300,7 +186,6 @@ pub async fn libpq_listener_main(
auth_type: AuthType,
tls_config: Option<Arc<rustls::ServerConfig>>,
pipelining_config: PageServicePipeliningConfig,
basebackup_cache: Arc<BasebackupCache>,
listener_ctx: RequestContext,
listener_cancel: CancellationToken,
) -> Connections {
@@ -344,7 +229,6 @@ pub async fn libpq_listener_main(
auth_type,
tls_config.clone(),
pipelining_config.clone(),
Arc::clone(&basebackup_cache),
connection_ctx,
connections_cancel.child_token(),
gate_guard,
@@ -370,7 +254,7 @@ type ConnectionHandlerResult = anyhow::Result<()>;
/// Perf root spans start at the per-request level, after shard routing.
/// This struct carries connection-level information to the root perf span definition.
#[derive(Clone, Default)]
#[derive(Clone)]
struct ConnectionPerfSpanFields {
peer_addr: String,
application_name: Option<String>,
@@ -387,7 +271,6 @@ async fn page_service_conn_main(
auth_type: AuthType,
tls_config: Option<Arc<rustls::ServerConfig>>,
pipelining_config: PageServicePipeliningConfig,
basebackup_cache: Arc<BasebackupCache>,
connection_ctx: RequestContext,
cancel: CancellationToken,
gate_guard: GateGuard,
@@ -448,12 +331,11 @@ async fn page_service_conn_main(
// But it's in a shared crate, so, we store connection_ctx inside PageServerHandler
// and create the per-query context in process_query ourselves.
let mut conn_handler = PageServerHandler::new(
conf,
tenant_manager,
auth,
pipelining_config,
conf.get_vectored_concurrent_io,
perf_span_fields,
basebackup_cache,
connection_ctx,
cancel.clone(),
gate_guard,
@@ -488,12 +370,8 @@ async fn page_service_conn_main(
}
}
/// Page service connection handler.
///
/// TODO: for gRPC, this will be shared by all requests from all connections.
/// Decompose it into global state and per-connection/request state, and make
/// libpq-specific options (e.g. pipelining) separate.
struct PageServerHandler {
conf: &'static PageServerConf,
auth: Option<Arc<SwappableJwtAuth>>,
claims: Option<Claims>,
@@ -511,9 +389,6 @@ struct PageServerHandler {
timeline_handles: Option<TimelineHandles>,
pipelining_config: PageServicePipeliningConfig,
get_vectored_concurrent_io: GetVectoredConcurrentIo,
basebackup_cache: Arc<BasebackupCache>,
gate_guard: GateGuard,
}
@@ -767,11 +642,8 @@ impl std::fmt::Display for BatchedPageStreamError {
struct BatchedGetPageRequest {
req: PagestreamGetPageRequest,
timer: SmgrOpTimer,
lsn_range: LsnRange,
effective_request_lsn: Lsn,
ctx: RequestContext,
// If the request is perf enabled, this contains a context
// with a perf span tracking the time spent waiting for the executor.
batch_wait_ctx: Option<RequestContext>,
}
#[cfg(feature = "testing")]
@@ -784,7 +656,6 @@ struct BatchedTestRequest {
/// so that we don't keep the [`Timeline::gate`] open while the batch
/// is being built up inside the [`spsc_fold`] (pagestream pipelining).
#[derive(IntoStaticStr)]
#[allow(clippy::large_enum_variant)]
enum BatchedFeMessage {
Exists {
span: Span,
@@ -893,12 +764,12 @@ impl BatchedFeMessage {
match batching_strategy {
PageServiceProtocolPipelinedBatchingStrategy::UniformLsn => {
if let Some(last_in_batch) = accum_pages.last() {
if last_in_batch.lsn_range.effective_lsn
!= this_pages[0].lsn_range.effective_lsn
if last_in_batch.effective_request_lsn
!= this_pages[0].effective_request_lsn
{
trace!(
accum_lsn = %last_in_batch.lsn_range.effective_lsn,
this_lsn = %this_pages[0].lsn_range.effective_lsn,
accum_lsn = %last_in_batch.effective_request_lsn,
this_lsn = %this_pages[0].effective_request_lsn,
"stopping batching because LSN changed"
);
@@ -913,15 +784,15 @@ impl BatchedFeMessage {
let same_page_different_lsn = accum_pages.iter().any(|batched| {
batched.req.rel == this_pages[0].req.rel
&& batched.req.blkno == this_pages[0].req.blkno
&& batched.lsn_range.effective_lsn
!= this_pages[0].lsn_range.effective_lsn
&& batched.effective_request_lsn
!= this_pages[0].effective_request_lsn
});
if same_page_different_lsn {
trace!(
rel=%this_pages[0].req.rel,
blkno=%this_pages[0].req.blkno,
lsn=%this_pages[0].lsn_range.effective_lsn,
lsn=%this_pages[0].effective_request_lsn,
"stopping batching because same page was requested at different LSNs"
);
@@ -973,17 +844,17 @@ impl BatchedFeMessage {
impl PageServerHandler {
#[allow(clippy::too_many_arguments)]
pub fn new(
conf: &'static PageServerConf,
tenant_manager: Arc<TenantManager>,
auth: Option<Arc<SwappableJwtAuth>>,
pipelining_config: PageServicePipeliningConfig,
get_vectored_concurrent_io: GetVectoredConcurrentIo,
perf_span_fields: ConnectionPerfSpanFields,
basebackup_cache: Arc<BasebackupCache>,
connection_ctx: RequestContext,
cancel: CancellationToken,
gate_guard: GateGuard,
) -> Self {
PageServerHandler {
conf,
auth,
claims: None,
connection_ctx,
@@ -991,8 +862,6 @@ impl PageServerHandler {
timeline_handles: Some(TimelineHandles::new(tenant_manager)),
cancel,
pipelining_config,
get_vectored_concurrent_io,
basebackup_cache,
gate_guard,
}
}
@@ -1289,7 +1158,7 @@ impl PageServerHandler {
.await?;
// We're holding the Handle
let effective_lsn = match Self::effective_request_lsn(
let effective_request_lsn = match Self::effective_request_lsn(
&shard,
shard.get_last_record_lsn(),
req.hdr.request_lsn,
@@ -1302,34 +1171,14 @@ impl PageServerHandler {
}
};
let batch_wait_ctx = if ctx.has_perf_span() {
Some(
RequestContextBuilder::from(&ctx)
.perf_span(|crnt_perf_span| {
info_span!(
target: PERF_TRACE_TARGET,
parent: crnt_perf_span,
"WAIT_EXECUTOR",
)
})
.attached_child(),
)
} else {
None
};
BatchedFeMessage::GetPage {
span,
shard: shard.downgrade(),
pages: smallvec::smallvec![BatchedGetPageRequest {
req,
timer,
lsn_range: LsnRange {
effective_lsn,
request_lsn: req.hdr.request_lsn
},
effective_request_lsn,
ctx,
batch_wait_ctx,
}],
// The executor grabs the batch when it becomes idle.
// Hence, [`GetPageBatchBreakReason::ExecutorSteal`] is the
@@ -1429,7 +1278,7 @@ impl PageServerHandler {
}
#[instrument(level = tracing::Level::DEBUG, skip_all)]
async fn pagestream_handle_batched_message<IO>(
async fn pagesteam_handle_batched_message<IO>(
&mut self,
pgb_writer: &mut PostgresBackend<IO>,
batch: BatchedFeMessage,
@@ -1485,7 +1334,7 @@ impl PageServerHandler {
let mut flush_timers = Vec::with_capacity(handler_results.len());
for handler_result in &mut handler_results {
let flush_timer = match handler_result {
Ok((_response, timer, _ctx)) => Some(
Ok((_, timer)) => Some(
timer
.observe_execution_end(flushing_start_time)
.expect("we are the first caller"),
@@ -1505,7 +1354,7 @@ impl PageServerHandler {
// Some handler errors cause exit from pagestream protocol.
// Other handler errors are sent back as an error message and we stay in pagestream protocol.
for (handler_result, flushing_timer) in handler_results.into_iter().zip(flush_timers) {
let (response_msg, ctx) = match handler_result {
let response_msg = match handler_result {
Err(e) => match &e.err {
PageStreamError::Shutdown => {
// If we fail to fulfil a request during shutdown, which may be _because_ of
@@ -1530,30 +1379,15 @@ impl PageServerHandler {
error!("error reading relation or page version: {full:#}")
});
(
PagestreamBeMessage::Error(PagestreamErrorResponse {
req: e.req,
message: e.err.to_string(),
}),
None,
)
PagestreamBeMessage::Error(PagestreamErrorResponse {
req: e.req,
message: e.err.to_string(),
})
}
},
Ok((response_msg, _op_timer_already_observed, ctx)) => (response_msg, Some(ctx)),
Ok((response_msg, _op_timer_already_observed)) => response_msg,
};
let ctx = ctx.map(|req_ctx| {
RequestContextBuilder::from(&req_ctx)
.perf_span(|crnt_perf_span| {
info_span!(
target: PERF_TRACE_TARGET,
parent: crnt_perf_span,
"FLUSH_RESPONSE",
)
})
.attached_child()
});
//
// marshal & transmit response message
//
@@ -1576,17 +1410,6 @@ impl PageServerHandler {
)),
None => futures::future::Either::Right(flush_fut),
};
let flush_fut = if let Some(req_ctx) = ctx.as_ref() {
futures::future::Either::Left(
flush_fut.maybe_perf_instrument(req_ctx, |current_perf_span| {
current_perf_span.clone()
}),
)
} else {
futures::future::Either::Right(flush_fut)
};
// do it while respecting cancellation
let _: () = async move {
tokio::select! {
@@ -1616,7 +1439,7 @@ impl PageServerHandler {
ctx: &RequestContext,
) -> Result<
(
Vec<Result<(PagestreamBeMessage, SmgrOpTimer, RequestContext), BatchedPageStreamError>>,
Vec<Result<(PagestreamBeMessage, SmgrOpTimer), BatchedPageStreamError>>,
Span,
),
QueryError,
@@ -1643,7 +1466,7 @@ impl PageServerHandler {
self.handle_get_rel_exists_request(&shard, &req, &ctx)
.instrument(span.clone())
.await
.map(|msg| (msg, timer, ctx))
.map(|msg| (msg, timer))
.map_err(|err| BatchedPageStreamError { err, req: req.hdr }),
],
span,
@@ -1662,7 +1485,7 @@ impl PageServerHandler {
self.handle_get_nblocks_request(&shard, &req, &ctx)
.instrument(span.clone())
.await
.map(|msg| (msg, timer, ctx))
.map(|msg| (msg, timer))
.map_err(|err| BatchedPageStreamError { err, req: req.hdr }),
],
span,
@@ -1709,7 +1532,7 @@ impl PageServerHandler {
self.handle_db_size_request(&shard, &req, &ctx)
.instrument(span.clone())
.await
.map(|msg| (msg, timer, ctx))
.map(|msg| (msg, timer))
.map_err(|err| BatchedPageStreamError { err, req: req.hdr }),
],
span,
@@ -1728,7 +1551,7 @@ impl PageServerHandler {
self.handle_get_slru_segment_request(&shard, &req, &ctx)
.instrument(span.clone())
.await
.map(|msg| (msg, timer, ctx))
.map(|msg| (msg, timer))
.map_err(|err| BatchedPageStreamError { err, req: req.hdr }),
],
span,
@@ -1800,7 +1623,7 @@ impl PageServerHandler {
}
let io_concurrency = IoConcurrency::spawn_from_conf(
self.get_vectored_concurrent_io,
self.conf,
match self.gate_guard.try_clone() {
Ok(guard) => guard,
Err(_) => {
@@ -1910,7 +1733,7 @@ impl PageServerHandler {
};
let result = self
.pagestream_handle_batched_message(
.pagesteam_handle_batched_message(
pgb_writer,
msg,
io_concurrency.clone(),
@@ -2080,26 +1903,13 @@ impl PageServerHandler {
return Ok(());
}
};
let mut batch = match batch {
let batch = match batch {
Ok(batch) => batch,
Err(e) => {
return Err(e);
}
};
if let BatchedFeMessage::GetPage {
pages,
span: _,
shard: _,
batch_break_reason: _,
} = &mut batch
{
for req in pages {
req.batch_wait_ctx.take();
}
}
self.pagestream_handle_batched_message(
self.pagesteam_handle_batched_message(
pgb_writer,
batch,
io_concurrency.clone(),
@@ -2317,14 +2127,7 @@ impl PageServerHandler {
.await?;
let exists = timeline
.get_rel_exists(
req.rel,
Version::LsnRange(LsnRange {
effective_lsn: lsn,
request_lsn: req.hdr.request_lsn,
}),
ctx,
)
.get_rel_exists(req.rel, Version::Lsn(lsn), ctx)
.await?;
Ok(PagestreamBeMessage::Exists(PagestreamExistsResponse {
@@ -2351,14 +2154,7 @@ impl PageServerHandler {
.await?;
let n_blocks = timeline
.get_rel_size(
req.rel,
Version::LsnRange(LsnRange {
effective_lsn: lsn,
request_lsn: req.hdr.request_lsn,
}),
ctx,
)
.get_rel_size(req.rel, Version::Lsn(lsn), ctx)
.await?;
Ok(PagestreamBeMessage::Nblocks(PagestreamNblocksResponse {
@@ -2385,15 +2181,7 @@ impl PageServerHandler {
.await?;
let total_blocks = timeline
.get_db_size(
DEFAULTTABLESPACE_OID,
req.dbnode,
Version::LsnRange(LsnRange {
effective_lsn: lsn,
request_lsn: req.hdr.request_lsn,
}),
ctx,
)
.get_db_size(DEFAULTTABLESPACE_OID, req.dbnode, Version::Lsn(lsn), ctx)
.await?;
let db_size = total_blocks as i64 * BLCKSZ as i64;
@@ -2411,8 +2199,7 @@ impl PageServerHandler {
io_concurrency: IoConcurrency,
batch_break_reason: GetPageBatchBreakReason,
ctx: &RequestContext,
) -> Vec<Result<(PagestreamBeMessage, SmgrOpTimer, RequestContext), BatchedPageStreamError>>
{
) -> Vec<Result<(PagestreamBeMessage, SmgrOpTimer), BatchedPageStreamError>> {
debug_assert_current_span_has_tenant_and_timeline_id();
timeline
@@ -2427,7 +2214,7 @@ impl PageServerHandler {
// Ignore error (trace buffer may be full or tracer may have disconnected).
_ = page_trace.try_send(PageTraceEvent {
key,
effective_lsn: batch.lsn_range.effective_lsn,
effective_lsn: batch.effective_request_lsn,
time,
});
}
@@ -2442,7 +2229,7 @@ impl PageServerHandler {
perf_instrument = true;
}
req.lsn_range.effective_lsn
req.effective_request_lsn
})
.max()
.expect("batch is never empty");
@@ -2496,7 +2283,7 @@ impl PageServerHandler {
(
&p.req.rel,
&p.req.blkno,
p.lsn_range,
p.effective_request_lsn,
p.ctx.attached_child(),
)
}),
@@ -2519,7 +2306,6 @@ impl PageServerHandler {
page,
}),
req.timer,
req.ctx,
)
})
.map_err(|e| BatchedPageStreamError {
@@ -2564,8 +2350,7 @@ impl PageServerHandler {
timeline: &Timeline,
requests: Vec<BatchedTestRequest>,
_ctx: &RequestContext,
) -> Vec<Result<(PagestreamBeMessage, SmgrOpTimer, RequestContext), BatchedPageStreamError>>
{
) -> Vec<Result<(PagestreamBeMessage, SmgrOpTimer), BatchedPageStreamError>> {
// real requests would do something with the timeline
let mut results = Vec::with_capacity(requests.len());
for _req in requests.iter() {
@@ -2592,10 +2377,6 @@ impl PageServerHandler {
req: req.req.clone(),
}),
req.timer,
RequestContext::new(
TaskKind::PageRequestHandler,
DownloadBehavior::Warn,
),
)
})
.map_err(|e| BatchedPageStreamError {
@@ -2687,8 +2468,6 @@ impl PageServerHandler {
.map_err(QueryError::Disconnected)?;
self.flush_cancellable(pgb, &self.cancel).await?;
let mut from_cache = false;
// Send a tarball of the latest layer on the timeline. Compress if not
// fullbackup. TODO Compress in that case too (tests need to be updated)
if full_backup {
@@ -2706,33 +2485,7 @@ impl PageServerHandler {
.map_err(map_basebackup_error)?;
} else {
let mut writer = BufWriter::new(pgb.copyout_writer());
let cached = {
// Basebackup is cached only for this combination of parameters.
if timeline.is_basebackup_cache_enabled()
&& gzip
&& lsn.is_some()
&& prev_lsn.is_none()
{
self.basebackup_cache
.get(tenant_id, timeline_id, lsn.unwrap())
.await
} else {
None
}
};
if let Some(mut cached) = cached {
from_cache = true;
tokio::io::copy(&mut cached, &mut writer)
.await
.map_err(|e| {
map_basebackup_error(BasebackupError::Client(
e,
"handle_basebackup_request,cached,copy",
))
})?;
} else if gzip {
if gzip {
let mut encoder = GzipEncoder::with_quality(
&mut writer,
// NOTE using fast compression because it's on the critical path
@@ -2791,7 +2544,6 @@ impl PageServerHandler {
info!(
lsn_await_millis = lsn_awaited_after.as_millis(),
basebackup_millis = basebackup_after.as_millis(),
%from_cache,
"basebackup complete"
);
@@ -3300,60 +3052,6 @@ where
}
}
/// Implements the page service over gRPC.
///
/// TODO: not yet implemented, all methods return unimplemented.
#[tonic::async_trait]
impl proto::PageService for PageServerHandler {
type GetBaseBackupStream = Pin<
Box<dyn Stream<Item = Result<proto::GetBaseBackupResponseChunk, tonic::Status>> + Send>,
>;
type GetPagesStream =
Pin<Box<dyn Stream<Item = Result<proto::GetPageResponse, tonic::Status>> + Send>>;
async fn check_rel_exists(
&self,
_: tonic::Request<proto::CheckRelExistsRequest>,
) -> Result<tonic::Response<proto::CheckRelExistsResponse>, tonic::Status> {
Err(tonic::Status::unimplemented("not implemented"))
}
async fn get_base_backup(
&self,
_: tonic::Request<proto::GetBaseBackupRequest>,
) -> Result<tonic::Response<Self::GetBaseBackupStream>, tonic::Status> {
Err(tonic::Status::unimplemented("not implemented"))
}
async fn get_db_size(
&self,
_: tonic::Request<proto::GetDbSizeRequest>,
) -> Result<tonic::Response<proto::GetDbSizeResponse>, tonic::Status> {
Err(tonic::Status::unimplemented("not implemented"))
}
async fn get_pages(
&self,
_: tonic::Request<tonic::Streaming<proto::GetPageRequest>>,
) -> Result<tonic::Response<Self::GetPagesStream>, tonic::Status> {
Err(tonic::Status::unimplemented("not implemented"))
}
async fn get_rel_size(
&self,
_: tonic::Request<proto::GetRelSizeRequest>,
) -> Result<tonic::Response<proto::GetRelSizeResponse>, tonic::Status> {
Err(tonic::Status::unimplemented("not implemented"))
}
async fn get_slru_segment(
&self,
_: tonic::Request<proto::GetSlruSegmentRequest>,
) -> Result<tonic::Response<proto::GetSlruSegmentResponse>, tonic::Status> {
Err(tonic::Status::unimplemented("not implemented"))
}
}
impl From<GetActiveTenantError> for QueryError {
fn from(e: GetActiveTenantError) -> Self {
match e {
@@ -3370,104 +3068,6 @@ impl From<GetActiveTenantError> for QueryError {
}
}
/// gRPC interceptor that decodes tenant metadata and stores it as request extensions of type
/// TenantTimelineId and ShardIndex.
///
/// TODO: consider looking up the timeline handle here and storing it.
#[derive(Clone)]
struct TenantMetadataInterceptor;
impl tonic::service::Interceptor for TenantMetadataInterceptor {
fn call(&mut self, mut req: tonic::Request<()>) -> Result<tonic::Request<()>, tonic::Status> {
// Decode the tenant ID.
let tenant_id = req
.metadata()
.get("neon-tenant-id")
.ok_or_else(|| tonic::Status::invalid_argument("missing neon-tenant-id"))?
.to_str()
.map_err(|_| tonic::Status::invalid_argument("invalid neon-tenant-id"))?;
let tenant_id = TenantId::from_str(tenant_id)
.map_err(|_| tonic::Status::invalid_argument("invalid neon-tenant-id"))?;
// Decode the timeline ID.
let timeline_id = req
.metadata()
.get("neon-timeline-id")
.ok_or_else(|| tonic::Status::invalid_argument("missing neon-timeline-id"))?
.to_str()
.map_err(|_| tonic::Status::invalid_argument("invalid neon-timeline-id"))?;
let timeline_id = TimelineId::from_str(timeline_id)
.map_err(|_| tonic::Status::invalid_argument("invalid neon-timeline-id"))?;
// Decode the shard ID.
let shard_index = req
.metadata()
.get("neon-shard-id")
.ok_or_else(|| tonic::Status::invalid_argument("missing neon-shard-id"))?
.to_str()
.map_err(|_| tonic::Status::invalid_argument("invalid neon-shard-id"))?;
let shard_index = ShardIndex::from_str(shard_index)
.map_err(|_| tonic::Status::invalid_argument("invalid neon-shard-id"))?;
// Stash them in the request.
let extensions = req.extensions_mut();
extensions.insert(TenantTimelineId::new(tenant_id, timeline_id));
extensions.insert(shard_index);
Ok(req)
}
}
/// Authenticates gRPC page service requests. Must run after TenantMetadataInterceptor.
#[derive(Clone)]
struct TenantAuthInterceptor {
auth: Option<Arc<SwappableJwtAuth>>,
}
impl TenantAuthInterceptor {
fn new(auth: Option<Arc<SwappableJwtAuth>>) -> Self {
Self { auth }
}
}
impl tonic::service::Interceptor for TenantAuthInterceptor {
fn call(&mut self, req: tonic::Request<()>) -> Result<tonic::Request<()>, tonic::Status> {
// Do nothing if auth is disabled.
let Some(auth) = self.auth.as_ref() else {
return Ok(req);
};
// Fetch the tenant ID that's been set by TenantMetadataInterceptor.
let ttid = req
.extensions()
.get::<TenantTimelineId>()
.expect("TenantMetadataInterceptor must run before TenantAuthInterceptor");
// Fetch and decode the JWT token.
let jwt = req
.metadata()
.get("authorization")
.ok_or_else(|| tonic::Status::unauthenticated("no authorization header"))?
.to_str()
.map_err(|_| tonic::Status::invalid_argument("invalid authorization header"))?
.strip_prefix("Bearer ")
.ok_or_else(|| tonic::Status::invalid_argument("invalid authorization header"))?
.trim();
let jwtdata: TokenData<Claims> = auth
.decode(jwt)
.map_err(|err| tonic::Status::invalid_argument(format!("invalid JWT token: {err}")))?;
let claims = jwtdata.claims;
// Check if the token is valid for this tenant.
check_permission(&claims, Some(ttid.tenant_id))
.map_err(|err| tonic::Status::permission_denied(err.to_string()))?;
// TODO: consider stashing the claims in the request extensions, if needed.
Ok(req)
}
}
#[derive(Debug, thiserror::Error)]
pub(crate) enum GetActiveTimelineError {
#[error(transparent)]

View File

@@ -43,9 +43,7 @@ use crate::aux_file;
use crate::context::{PerfInstrumentFutureExt, RequestContext, RequestContextBuilder};
use crate::keyspace::{KeySpace, KeySpaceAccum};
use crate::metrics::{
RELSIZE_CACHE_MISSES_OLD, RELSIZE_LATEST_CACHE_ENTRIES, RELSIZE_LATEST_CACHE_HITS,
RELSIZE_LATEST_CACHE_MISSES, RELSIZE_SNAPSHOT_CACHE_ENTRIES, RELSIZE_SNAPSHOT_CACHE_HITS,
RELSIZE_SNAPSHOT_CACHE_MISSES,
RELSIZE_CACHE_ENTRIES, RELSIZE_CACHE_HITS, RELSIZE_CACHE_MISSES, RELSIZE_CACHE_MISSES_OLD,
};
use crate::span::{
debug_assert_current_span_has_tenant_and_timeline_id,
@@ -92,28 +90,6 @@ pub enum LsnForTimestamp {
NoData(Lsn),
}
/// Each request to page server contains LSN range: `not_modified_since..request_lsn`.
/// See comments libs/pageserver_api/src/models.rs.
/// Based on this range and `last_record_lsn` PS calculates `effective_lsn`.
/// But to distinguish requests from primary and replicas we need also to pass `request_lsn`.
#[derive(Debug, Clone, Copy, Default)]
pub struct LsnRange {
pub effective_lsn: Lsn,
pub request_lsn: Lsn,
}
impl LsnRange {
pub fn at(lsn: Lsn) -> LsnRange {
LsnRange {
effective_lsn: lsn,
request_lsn: lsn,
}
}
pub fn is_latest(&self) -> bool {
self.request_lsn == Lsn::MAX
}
}
#[derive(Debug, thiserror::Error)]
pub(crate) enum CalculateLogicalSizeError {
#[error("cancelled")]
@@ -226,13 +202,13 @@ impl Timeline {
io_concurrency: IoConcurrency,
) -> Result<Bytes, PageReconstructError> {
match version {
Version::LsnRange(lsns) => {
Version::Lsn(effective_lsn) => {
let pages: smallvec::SmallVec<[_; 1]> = smallvec::smallvec![(tag, blknum)];
let res = self
.get_rel_page_at_lsn_batched(
pages
.iter()
.map(|(tag, blknum)| (tag, blknum, lsns, ctx.attached_child())),
pages.iter().map(|(tag, blknum)| {
(tag, blknum, effective_lsn, ctx.attached_child())
}),
io_concurrency.clone(),
ctx,
)
@@ -270,7 +246,7 @@ impl Timeline {
/// The ordering of the returned vec corresponds to the ordering of `pages`.
pub(crate) async fn get_rel_page_at_lsn_batched(
&self,
pages: impl ExactSizeIterator<Item = (&RelTag, &BlockNumber, LsnRange, RequestContext)>,
pages: impl ExactSizeIterator<Item = (&RelTag, &BlockNumber, Lsn, RequestContext)>,
io_concurrency: IoConcurrency,
ctx: &RequestContext,
) -> Vec<Result<Bytes, PageReconstructError>> {
@@ -289,7 +265,7 @@ impl Timeline {
let mut req_keyspaces: HashMap<Lsn, KeySpaceRandomAccum> =
HashMap::with_capacity(pages.len());
for (response_slot_idx, (tag, blknum, lsns, ctx)) in pages.enumerate() {
for (response_slot_idx, (tag, blknum, lsn, ctx)) in pages.enumerate() {
if tag.relnode == 0 {
result_slots[response_slot_idx].write(Err(PageReconstructError::Other(
RelationError::InvalidRelnode.into(),
@@ -298,7 +274,7 @@ impl Timeline {
slots_filled += 1;
continue;
}
let lsn = lsns.effective_lsn;
let nblocks = {
let ctx = RequestContextBuilder::from(&ctx)
.perf_span(|crnt_perf_span| {
@@ -313,7 +289,7 @@ impl Timeline {
.attached_child();
match self
.get_rel_size(*tag, Version::LsnRange(lsns), &ctx)
.get_rel_size(*tag, Version::Lsn(lsn), &ctx)
.maybe_perf_instrument(&ctx, |crnt_perf_span| crnt_perf_span.clone())
.await
{
@@ -494,7 +470,7 @@ impl Timeline {
));
}
if let Some(nblocks) = self.get_cached_rel_size(&tag, version) {
if let Some(nblocks) = self.get_cached_rel_size(&tag, version.get_lsn()) {
return Ok(nblocks);
}
@@ -512,7 +488,7 @@ impl Timeline {
let mut buf = version.get(self, key, ctx).await?;
let nblocks = buf.get_u32_le();
self.update_cached_rel_size(tag, version, nblocks);
self.update_cached_rel_size(tag, version.get_lsn(), nblocks);
Ok(nblocks)
}
@@ -534,7 +510,7 @@ impl Timeline {
}
// first try to lookup relation in cache
if let Some(_nblocks) = self.get_cached_rel_size(&tag, version) {
if let Some(_nblocks) = self.get_cached_rel_size(&tag, version.get_lsn()) {
return Ok(true);
}
// then check if the database was already initialized.
@@ -610,7 +586,7 @@ impl Timeline {
// scan directory listing (new), merge with the old results
let key_range = rel_tag_sparse_key_range(spcnode, dbnode);
let io_concurrency = IoConcurrency::spawn_from_conf(
self.conf.get_vectored_concurrent_io,
self.conf,
self.gate
.enter()
.map_err(|_| PageReconstructError::Cancelled)?,
@@ -656,7 +632,7 @@ impl Timeline {
) -> Result<Bytes, PageReconstructError> {
assert!(self.tenant_shard_id.is_shard_zero());
let n_blocks = self
.get_slru_segment_size(kind, segno, Version::at(lsn), ctx)
.get_slru_segment_size(kind, segno, Version::Lsn(lsn), ctx)
.await?;
let keyspace = KeySpace::single(
@@ -669,7 +645,7 @@ impl Timeline {
);
let io_concurrency = IoConcurrency::spawn_from_conf(
self.conf.get_vectored_concurrent_io,
self.conf,
self.gate
.enter()
.map_err(|_| PageReconstructError::Cancelled)?,
@@ -891,11 +867,11 @@ impl Timeline {
mut f: impl FnMut(TimestampTz) -> ControlFlow<T>,
) -> Result<T, PageReconstructError> {
for segno in self
.list_slru_segments(SlruKind::Clog, Version::at(probe_lsn), ctx)
.list_slru_segments(SlruKind::Clog, Version::Lsn(probe_lsn), ctx)
.await?
{
let nblocks = self
.get_slru_segment_size(SlruKind::Clog, segno, Version::at(probe_lsn), ctx)
.get_slru_segment_size(SlruKind::Clog, segno, Version::Lsn(probe_lsn), ctx)
.await?;
let keyspace = KeySpace::single(
@@ -909,7 +885,7 @@ impl Timeline {
);
let io_concurrency = IoConcurrency::spawn_from_conf(
self.conf.get_vectored_concurrent_io,
self.conf,
self.gate
.enter()
.map_err(|_| PageReconstructError::Cancelled)?,
@@ -1161,7 +1137,7 @@ impl Timeline {
let mut total_size: u64 = 0;
for (spcnode, dbnode) in dbdir.dbdirs.keys() {
for rel in self
.list_rels(*spcnode, *dbnode, Version::at(lsn), ctx)
.list_rels(*spcnode, *dbnode, Version::Lsn(lsn), ctx)
.await?
{
if self.cancel.is_cancelled() {
@@ -1236,7 +1212,7 @@ impl Timeline {
result.add_key(rel_dir_to_key(spcnode, dbnode));
let mut rels: Vec<RelTag> = self
.list_rels(spcnode, dbnode, Version::at(lsn), ctx)
.list_rels(spcnode, dbnode, Version::Lsn(lsn), ctx)
.await?
.into_iter()
.collect();
@@ -1353,75 +1329,59 @@ impl Timeline {
Ok((dense_keyspace, sparse_keyspace))
}
/// Get cached size of relation. There are two caches: one for primary updates, it captures the latest state of
/// of the timeline and snapshot cache, which key includes LSN and so can be used by replicas to get relation size
/// at the particular LSN (snapshot).
pub fn get_cached_rel_size(&self, tag: &RelTag, version: Version<'_>) -> Option<BlockNumber> {
let lsn = version.get_lsn();
{
let rel_size_cache = self.rel_size_latest_cache.read().unwrap();
if let Some((cached_lsn, nblocks)) = rel_size_cache.get(tag) {
if lsn >= *cached_lsn {
RELSIZE_LATEST_CACHE_HITS.inc();
return Some(*nblocks);
}
RELSIZE_CACHE_MISSES_OLD.inc();
/// Get cached size of relation if it not updated after specified LSN
pub fn get_cached_rel_size(&self, tag: &RelTag, lsn: Lsn) -> Option<BlockNumber> {
let rel_size_cache = self.rel_size_cache.read().unwrap();
if let Some((cached_lsn, nblocks)) = rel_size_cache.map.get(tag) {
if lsn >= *cached_lsn {
RELSIZE_CACHE_HITS.inc();
return Some(*nblocks);
}
RELSIZE_CACHE_MISSES_OLD.inc();
}
{
let mut rel_size_cache = self.rel_size_snapshot_cache.lock().unwrap();
if let Some(nblock) = rel_size_cache.get(&(lsn, *tag)) {
RELSIZE_SNAPSHOT_CACHE_HITS.inc();
return Some(*nblock);
}
}
if version.is_latest() {
RELSIZE_LATEST_CACHE_MISSES.inc();
} else {
RELSIZE_SNAPSHOT_CACHE_MISSES.inc();
}
RELSIZE_CACHE_MISSES.inc();
None
}
/// Update cached relation size if there is no more recent update
pub fn update_cached_rel_size(&self, tag: RelTag, version: Version<'_>, nblocks: BlockNumber) {
let lsn = version.get_lsn();
if version.is_latest() {
let mut rel_size_cache = self.rel_size_latest_cache.write().unwrap();
match rel_size_cache.entry(tag) {
hash_map::Entry::Occupied(mut entry) => {
let cached_lsn = entry.get_mut();
if lsn >= cached_lsn.0 {
*cached_lsn = (lsn, nblocks);
}
}
hash_map::Entry::Vacant(entry) => {
entry.insert((lsn, nblocks));
RELSIZE_LATEST_CACHE_ENTRIES.inc();
pub fn update_cached_rel_size(&self, tag: RelTag, lsn: Lsn, nblocks: BlockNumber) {
let mut rel_size_cache = self.rel_size_cache.write().unwrap();
if lsn < rel_size_cache.complete_as_of {
// Do not cache old values. It's safe to cache the size on read, as long as
// the read was at an LSN since we started the WAL ingestion. Reasoning: we
// never evict values from the cache, so if the relation size changed after
// 'lsn', the new value is already in the cache.
return;
}
match rel_size_cache.map.entry(tag) {
hash_map::Entry::Occupied(mut entry) => {
let cached_lsn = entry.get_mut();
if lsn >= cached_lsn.0 {
*cached_lsn = (lsn, nblocks);
}
}
} else {
let mut rel_size_cache = self.rel_size_snapshot_cache.lock().unwrap();
if rel_size_cache.capacity() != 0 {
rel_size_cache.insert((lsn, tag), nblocks);
RELSIZE_SNAPSHOT_CACHE_ENTRIES.set(rel_size_cache.len() as u64);
hash_map::Entry::Vacant(entry) => {
entry.insert((lsn, nblocks));
RELSIZE_CACHE_ENTRIES.inc();
}
}
}
/// Store cached relation size
pub fn set_cached_rel_size(&self, tag: RelTag, lsn: Lsn, nblocks: BlockNumber) {
let mut rel_size_cache = self.rel_size_latest_cache.write().unwrap();
if rel_size_cache.insert(tag, (lsn, nblocks)).is_none() {
RELSIZE_LATEST_CACHE_ENTRIES.inc();
let mut rel_size_cache = self.rel_size_cache.write().unwrap();
if rel_size_cache.map.insert(tag, (lsn, nblocks)).is_none() {
RELSIZE_CACHE_ENTRIES.inc();
}
}
/// Remove cached relation size
pub fn remove_cached_rel_size(&self, tag: &RelTag) {
let mut rel_size_cache = self.rel_size_latest_cache.write().unwrap();
if rel_size_cache.remove(tag).is_some() {
RELSIZE_LATEST_CACHE_ENTRIES.dec();
let mut rel_size_cache = self.rel_size_cache.write().unwrap();
if rel_size_cache.map.remove(tag).is_some() {
RELSIZE_CACHE_ENTRIES.dec();
}
}
}
@@ -1625,10 +1585,7 @@ impl DatadirModification<'_> {
// check the cache too. This is because eagerly checking the cache results in
// less work overall and 10% better performance. It's more work on cache miss
// but cache miss is rare.
if let Some(nblocks) = self
.tline
.get_cached_rel_size(&rel, Version::Modified(self))
{
if let Some(nblocks) = self.tline.get_cached_rel_size(&rel, self.get_lsn()) {
Ok(nblocks)
} else if !self
.tline
@@ -2710,7 +2667,7 @@ pub struct DatadirModificationStats {
/// timeline to not miss the latest updates.
#[derive(Clone, Copy)]
pub enum Version<'a> {
LsnRange(LsnRange),
Lsn(Lsn),
Modified(&'a DatadirModification<'a>),
}
@@ -2722,7 +2679,7 @@ impl Version<'_> {
ctx: &RequestContext,
) -> Result<Bytes, PageReconstructError> {
match self {
Version::LsnRange(lsns) => timeline.get(key, lsns.effective_lsn, ctx).await,
Version::Lsn(lsn) => timeline.get(key, *lsn, ctx).await,
Version::Modified(modification) => modification.get(key, ctx).await,
}
}
@@ -2744,26 +2701,12 @@ impl Version<'_> {
}
}
pub fn is_latest(&self) -> bool {
fn get_lsn(&self) -> Lsn {
match self {
Version::LsnRange(lsns) => lsns.is_latest(),
Version::Modified(_) => true,
}
}
pub fn get_lsn(&self) -> Lsn {
match self {
Version::LsnRange(lsns) => lsns.effective_lsn,
Version::Lsn(lsn) => *lsn,
Version::Modified(modification) => modification.lsn,
}
}
pub fn at(lsn: Lsn) -> Self {
Version::LsnRange(LsnRange {
effective_lsn: lsn,
request_lsn: lsn,
})
}
}
//--- Metadata structs stored in key-value pairs in the repository.

View File

@@ -276,10 +276,9 @@ pub enum TaskKind {
// HTTP endpoint listener.
HttpEndpointListener,
/// Task that handles a single page service connection. A PageRequestHandler
/// task starts detached from any particular tenant or timeline, but it can
/// be associated with one later, after receiving a command from the client.
/// Also used for the gRPC page service API, including the main server task.
// Task that handles a single connection. A PageRequestHandler task
// starts detached from any particular tenant or timeline, but it can be
// associated with one later, after receiving a command from the client.
PageRequestHandler,
/// Manages the WAL receiver connection for one timeline.
@@ -381,10 +380,6 @@ pub enum TaskKind {
DetachAncestor,
ImportPgdata,
/// Background task of [`crate::basebackup_cache::BasebackupCache`].
/// Prepares basebackups and clears outdated entries.
BasebackupCache,
}
#[derive(Default)]

View File

@@ -78,18 +78,16 @@ use self::timeline::uninit::{TimelineCreateGuard, TimelineExclusionError, Uninit
use self::timeline::{
EvictionTaskTenantState, GcCutoffs, TimelineDeleteProgress, TimelineResources, WaitLsnError,
};
use crate::basebackup_cache::BasebackupPrepareSender;
use crate::config::PageServerConf;
use crate::context;
use crate::context::RequestContextBuilder;
use crate::context::{DownloadBehavior, RequestContext};
use crate::deletion_queue::{DeletionQueueClient, DeletionQueueError};
use crate::feature_resolver::FeatureResolver;
use crate::l0_flush::L0FlushGlobalState;
use crate::metrics::{
BROKEN_TENANTS_SET, CIRCUIT_BREAKERS_BROKEN, CIRCUIT_BREAKERS_UNBROKEN, CONCURRENT_INITDBS,
INITDB_RUN_TIME, INITDB_SEMAPHORE_ACQUISITION_TIME, TENANT, TENANT_OFFLOADED_TIMELINES,
TENANT_STATE_METRIC, TENANT_SYNTHETIC_SIZE_METRIC, remove_tenant_metrics,
INITDB_RUN_TIME, INITDB_SEMAPHORE_ACQUISITION_TIME, TENANT, TENANT_STATE_METRIC,
TENANT_SYNTHETIC_SIZE_METRIC, remove_tenant_metrics,
};
use crate::task_mgr::TaskKind;
use crate::tenant::config::LocationMode;
@@ -159,8 +157,6 @@ pub struct TenantSharedResources {
pub remote_storage: GenericRemoteStorage,
pub deletion_queue_client: DeletionQueueClient,
pub l0_flush_global_state: L0FlushGlobalState,
pub basebackup_prepare_sender: BasebackupPrepareSender,
pub feature_resolver: FeatureResolver,
}
/// A [`TenantShard`] is really an _attached_ tenant. The configuration
@@ -300,7 +296,7 @@ pub struct TenantShard {
/// as in progress.
/// * Imported timelines are removed when the storage controller calls the post timeline
/// import activation endpoint.
timelines_importing: std::sync::Mutex<HashMap<TimelineId, Arc<ImportingTimeline>>>,
timelines_importing: std::sync::Mutex<HashMap<TimelineId, ImportingTimeline>>,
/// The last tenant manifest known to be in remote storage. None if the manifest has not yet
/// been either downloaded or uploaded. Always Some after tenant attach.
@@ -321,15 +317,12 @@ pub struct TenantShard {
gc_cs: tokio::sync::Mutex<()>,
walredo_mgr: Option<Arc<WalRedoManager>>,
/// Provides access to timeline data sitting in the remote storage.
// provides access to timeline data sitting in the remote storage
pub(crate) remote_storage: GenericRemoteStorage,
/// Access to global deletion queue for when this tenant wants to schedule a deletion.
// Access to global deletion queue for when this tenant wants to schedule a deletion
deletion_queue_client: DeletionQueueClient,
/// A channel to send async requests to prepare a basebackup for the basebackup cache.
basebackup_prepare_sender: BasebackupPrepareSender,
/// Cached logical sizes updated updated on each [`TenantShard::gather_size_inputs`].
cached_logical_sizes: tokio::sync::Mutex<HashMap<(TimelineId, Lsn), u64>>,
cached_synthetic_tenant_size: Arc<AtomicU64>,
@@ -382,8 +375,6 @@ pub struct TenantShard {
pub(crate) gc_block: gc_block::GcBlock,
l0_flush_global_state: L0FlushGlobalState,
pub(crate) feature_resolver: FeatureResolver,
}
impl std::fmt::Debug for TenantShard {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
@@ -672,7 +663,6 @@ pub enum MaybeOffloaded {
pub enum TimelineOrOffloaded {
Timeline(Arc<Timeline>),
Offloaded(Arc<OffloadedTimeline>),
Importing(Arc<ImportingTimeline>),
}
impl TimelineOrOffloaded {
@@ -684,9 +674,6 @@ impl TimelineOrOffloaded {
TimelineOrOffloaded::Offloaded(offloaded) => {
TimelineOrOffloadedArcRef::Offloaded(offloaded)
}
TimelineOrOffloaded::Importing(importing) => {
TimelineOrOffloadedArcRef::Importing(importing)
}
}
}
pub fn tenant_shard_id(&self) -> TenantShardId {
@@ -699,16 +686,12 @@ impl TimelineOrOffloaded {
match self {
TimelineOrOffloaded::Timeline(timeline) => &timeline.delete_progress,
TimelineOrOffloaded::Offloaded(offloaded) => &offloaded.delete_progress,
TimelineOrOffloaded::Importing(importing) => &importing.delete_progress,
}
}
fn maybe_remote_client(&self) -> Option<Arc<RemoteTimelineClient>> {
match self {
TimelineOrOffloaded::Timeline(timeline) => Some(timeline.remote_client.clone()),
TimelineOrOffloaded::Offloaded(_offloaded) => None,
TimelineOrOffloaded::Importing(importing) => {
Some(importing.timeline.remote_client.clone())
}
}
}
}
@@ -716,7 +699,6 @@ impl TimelineOrOffloaded {
pub enum TimelineOrOffloadedArcRef<'a> {
Timeline(&'a Arc<Timeline>),
Offloaded(&'a Arc<OffloadedTimeline>),
Importing(&'a Arc<ImportingTimeline>),
}
impl TimelineOrOffloadedArcRef<'_> {
@@ -724,14 +706,12 @@ impl TimelineOrOffloadedArcRef<'_> {
match self {
TimelineOrOffloadedArcRef::Timeline(timeline) => timeline.tenant_shard_id,
TimelineOrOffloadedArcRef::Offloaded(offloaded) => offloaded.tenant_shard_id,
TimelineOrOffloadedArcRef::Importing(importing) => importing.timeline.tenant_shard_id,
}
}
pub fn timeline_id(&self) -> TimelineId {
match self {
TimelineOrOffloadedArcRef::Timeline(timeline) => timeline.timeline_id,
TimelineOrOffloadedArcRef::Offloaded(offloaded) => offloaded.timeline_id,
TimelineOrOffloadedArcRef::Importing(importing) => importing.timeline.timeline_id,
}
}
}
@@ -748,12 +728,6 @@ impl<'a> From<&'a Arc<OffloadedTimeline>> for TimelineOrOffloadedArcRef<'a> {
}
}
impl<'a> From<&'a Arc<ImportingTimeline>> for TimelineOrOffloadedArcRef<'a> {
fn from(timeline: &'a Arc<ImportingTimeline>) -> Self {
Self::Importing(timeline)
}
}
#[derive(Debug, thiserror::Error, PartialEq, Eq)]
pub enum GetTimelineError {
#[error("Timeline is shutting down")]
@@ -881,14 +855,6 @@ impl Debug for SetStoppingError {
}
}
#[derive(thiserror::Error, Debug)]
pub(crate) enum FinalizeTimelineImportError {
#[error("Import task not done yet")]
ImportTaskStillRunning,
#[error("Shutting down")]
ShuttingDown,
}
/// Arguments to [`TenantShard::create_timeline`].
///
/// Not usable as an idempotency key for timeline creation because if [`CreateTimelineParamsBranch::ancestor_start_lsn`]
@@ -1175,20 +1141,10 @@ impl TenantShard {
ctx,
)?;
let disk_consistent_lsn = timeline.get_disk_consistent_lsn();
if !disk_consistent_lsn.is_valid() {
// As opposed to normal timelines which get initialised with a disk consitent LSN
// via initdb, imported timelines start from 0. If the import task stops before
// it advances disk consitent LSN, allow it to resume.
let in_progress_import = import_pgdata
.as_ref()
.map(|import| !import.is_done())
.unwrap_or(false);
if !in_progress_import {
anyhow::bail!("Timeline {tenant_id}/{timeline_id} has invalid disk_consistent_lsn");
}
}
anyhow::ensure!(
disk_consistent_lsn.is_valid(),
"Timeline {tenant_id}/{timeline_id} has invalid disk_consistent_lsn"
);
assert_eq!(
disk_consistent_lsn,
metadata.disk_consistent_lsn(),
@@ -1282,25 +1238,20 @@ impl TenantShard {
}
}
if disk_consistent_lsn.is_valid() {
// Sanity check: a timeline should have some content.
// Exception: importing timelines might not yet have any
anyhow::ensure!(
ancestor.is_some()
|| timeline
.layers
.read()
.await
.layer_map()
.expect(
"currently loading, layer manager cannot be shutdown already"
)
.iter_historic_layers()
.next()
.is_some(),
"Timeline has no ancestor and no layer files"
);
}
// Sanity check: a timeline should have some content.
anyhow::ensure!(
ancestor.is_some()
|| timeline
.layers
.read()
.await
.layer_map()
.expect("currently loading, layer manager cannot be shutdown already")
.iter_historic_layers()
.next()
.is_some(),
"Timeline has no ancestor and no layer files"
);
Ok(TimelineInitAndSyncResult::ReadyToActivate)
}
@@ -1335,8 +1286,6 @@ impl TenantShard {
remote_storage,
deletion_queue_client,
l0_flush_global_state,
basebackup_prepare_sender,
feature_resolver,
} = resources;
let attach_mode = attached_conf.location.attach_mode;
@@ -1352,8 +1301,6 @@ impl TenantShard {
remote_storage.clone(),
deletion_queue_client,
l0_flush_global_state,
basebackup_prepare_sender,
feature_resolver,
));
// The attach task will carry a GateGuard, so that shutdown() reliably waits for it to drop out if
@@ -1806,25 +1753,20 @@ impl TenantShard {
},
) => {
let timeline_id = timeline.timeline_id;
let import_task_gate = Gate::default();
let import_task_guard = import_task_gate.enter().unwrap();
let import_task_handle =
tokio::task::spawn(self.clone().create_timeline_import_pgdata_task(
timeline.clone(),
import_pgdata,
guard,
import_task_guard,
ctx.detached_child(TaskKind::ImportPgdata, DownloadBehavior::Warn),
));
let prev = self.timelines_importing.lock().unwrap().insert(
timeline_id,
Arc::new(ImportingTimeline {
ImportingTimeline {
timeline: timeline.clone(),
import_task_handle,
import_task_gate,
delete_progress: TimelineDeleteProgress::default(),
}),
},
);
assert!(prev.is_none());
@@ -2442,17 +2384,6 @@ impl TenantShard {
.collect()
}
/// Lists timelines the tenant contains.
/// It's up to callers to omit certain timelines that are not considered ready for use.
pub fn list_importing_timelines(&self) -> Vec<Arc<ImportingTimeline>> {
self.timelines_importing
.lock()
.unwrap()
.values()
.map(Arc::clone)
.collect()
}
/// Lists timelines the tenant manages, including offloaded ones.
///
/// It's up to callers to omit certain timelines that are not considered ready for use.
@@ -2886,25 +2817,19 @@ impl TenantShard {
let (timeline, timeline_create_guard) = uninit_timeline.finish_creation_myself();
let import_task_gate = Gate::default();
let import_task_guard = import_task_gate.enter().unwrap();
let import_task_handle = tokio::spawn(self.clone().create_timeline_import_pgdata_task(
timeline.clone(),
index_part,
timeline_create_guard,
import_task_guard,
timeline_ctx.detached_child(TaskKind::ImportPgdata, DownloadBehavior::Warn),
));
let prev = self.timelines_importing.lock().unwrap().insert(
timeline.timeline_id,
Arc::new(ImportingTimeline {
ImportingTimeline {
timeline: timeline.clone(),
import_task_handle,
import_task_gate,
delete_progress: TimelineDeleteProgress::default(),
}),
},
);
// Idempotency is enforced higher up the stack
@@ -2922,13 +2847,13 @@ impl TenantShard {
pub(crate) async fn finalize_importing_timeline(
&self,
timeline_id: TimelineId,
) -> Result<(), FinalizeTimelineImportError> {
) -> anyhow::Result<()> {
let timeline = {
let locked = self.timelines_importing.lock().unwrap();
match locked.get(&timeline_id) {
Some(importing_timeline) => {
if !importing_timeline.import_task_handle.is_finished() {
return Err(FinalizeTimelineImportError::ImportTaskStillRunning);
return Err(anyhow::anyhow!("Import task not done yet"));
}
importing_timeline.timeline.clone()
@@ -2941,13 +2866,8 @@ impl TenantShard {
timeline
.remote_client
.schedule_index_upload_for_import_pgdata_finalize()
.map_err(|_err| FinalizeTimelineImportError::ShuttingDown)?;
timeline
.remote_client
.wait_completion()
.await
.map_err(|_err| FinalizeTimelineImportError::ShuttingDown)?;
.schedule_index_upload_for_import_pgdata_finalize()?;
timeline.remote_client.wait_completion().await?;
self.timelines_importing
.lock()
@@ -2963,7 +2883,6 @@ impl TenantShard {
timeline: Arc<Timeline>,
index_part: import_pgdata::index_part_format::Root,
timeline_create_guard: TimelineCreateGuard,
_import_task_guard: GateGuard,
ctx: RequestContext,
) {
debug_assert_current_span_has_tenant_and_timeline_id();
@@ -3209,18 +3128,11 @@ impl TenantShard {
.or_insert_with(|| Arc::new(GcCompactionQueue::new()))
.clone()
};
let gc_compaction_strategy = self
.feature_resolver
.evaluate_multivariate("gc-comapction-strategy", self.tenant_shard_id.tenant_id)
.ok();
let span = if let Some(gc_compaction_strategy) = gc_compaction_strategy {
info_span!("gc_compact_timeline", timeline_id = %timeline.timeline_id, strategy = %gc_compaction_strategy)
} else {
info_span!("gc_compact_timeline", timeline_id = %timeline.timeline_id)
};
outcome = queue
.iteration(cancel, ctx, &self.gc_block, &timeline)
.instrument(span)
.instrument(
info_span!("gc_compact_timeline", timeline_id = %timeline.timeline_id),
)
.await?;
}
@@ -3436,13 +3348,6 @@ impl TenantShard {
activated_timelines += 1;
}
let tid = self.tenant_shard_id.tenant_id.to_string();
let shard_id = self.tenant_shard_id.shard_slug().to_string();
let offloaded_timeline_count = timelines_offloaded_accessor.len();
TENANT_OFFLOADED_TIMELINES
.with_label_values(&[&tid, &shard_id])
.set(offloaded_timeline_count as u64);
self.state.send_modify(move |current_state| {
assert!(
matches!(current_state, TenantState::Activating(_)),
@@ -3552,9 +3457,8 @@ impl TenantShard {
let mut timelines_importing = self.timelines_importing.lock().unwrap();
timelines_importing
.drain()
.for_each(|(timeline_id, importing_timeline)| {
let span = tracing::info_span!("importing_timeline_shutdown", %timeline_id);
js.spawn(async move { importing_timeline.shutdown().instrument(span).await });
.for_each(|(_timeline_id, importing_timeline)| {
importing_timeline.shutdown();
});
}
// test_long_timeline_create_then_tenant_delete is leaning on this message
@@ -3875,9 +3779,6 @@ impl TenantShard {
.build_timeline_client(offloaded.timeline_id, self.remote_storage.clone());
Arc::new(remote_client)
}
TimelineOrOffloadedArcRef::Importing(_) => {
unreachable!("Importing timelines are not included in the iterator")
}
};
// Shut down the timeline's remote client: this means that the indices we write
@@ -4331,8 +4232,6 @@ impl TenantShard {
remote_storage: GenericRemoteStorage,
deletion_queue_client: DeletionQueueClient,
l0_flush_global_state: L0FlushGlobalState,
basebackup_prepare_sender: BasebackupPrepareSender,
feature_resolver: FeatureResolver,
) -> TenantShard {
assert!(!attached_conf.location.generation.is_none());
@@ -4436,8 +4335,6 @@ impl TenantShard {
ongoing_timeline_detach: std::sync::Mutex::default(),
gc_block: Default::default(),
l0_flush_global_state,
basebackup_prepare_sender,
feature_resolver,
}
}
@@ -4690,7 +4587,7 @@ impl TenantShard {
target.cutoffs = GcCutoffs {
space: space_cutoff,
time: None,
time: Lsn::INVALID,
};
}
}
@@ -4774,7 +4671,7 @@ impl TenantShard {
if let Some(ancestor_id) = timeline.get_ancestor_timeline_id() {
if let Some(ancestor_gc_cutoffs) = gc_cutoffs.get(&ancestor_id) {
target.within_ancestor_pitr =
Some(timeline.get_ancestor_lsn()) >= ancestor_gc_cutoffs.time;
timeline.get_ancestor_lsn() >= ancestor_gc_cutoffs.time;
}
}
@@ -4787,15 +4684,13 @@ impl TenantShard {
} else {
0
});
if let Some(time_cutoff) = target.cutoffs.time {
timeline.metrics.pitr_history_size.set(
timeline
.get_last_record_lsn()
.checked_sub(time_cutoff)
.unwrap_or_default()
.0,
);
}
timeline.metrics.pitr_history_size.set(
timeline
.get_last_record_lsn()
.checked_sub(target.cutoffs.time)
.unwrap_or(Lsn(0))
.0,
);
// Apply the cutoffs we found to the Timeline's GcInfo. Why might we _not_ have cutoffs for a timeline?
// - this timeline was created while we were finding cutoffs
@@ -4804,8 +4699,8 @@ impl TenantShard {
let original_cutoffs = target.cutoffs.clone();
// GC cutoffs should never go back
target.cutoffs = GcCutoffs {
space: cutoffs.space.max(original_cutoffs.space),
time: cutoffs.time.max(original_cutoffs.time),
space: Lsn(cutoffs.space.0.max(original_cutoffs.space.0)),
time: Lsn(cutoffs.time.0.max(original_cutoffs.time.0)),
}
}
}
@@ -5087,14 +4982,6 @@ impl TenantShard {
info!("timeline already exists but is offloaded");
Err(CreateTimelineError::Conflict)
}
Err(TimelineExclusionError::AlreadyExists {
existing: TimelineOrOffloaded::Importing(_existing),
..
}) => {
// If there's a timeline already importing, then we would hit
// the [`TimelineExclusionError::AlreadyCreating`] branch above.
unreachable!("Importing timelines hold the creation guard")
}
Err(TimelineExclusionError::AlreadyExists {
existing: TimelineOrOffloaded::Timeline(existing),
arg,
@@ -5365,8 +5252,6 @@ impl TenantShard {
pagestream_throttle_metrics: self.pagestream_throttle_metrics.clone(),
l0_compaction_trigger: self.l0_compaction_trigger.clone(),
l0_flush_global_state: self.l0_flush_global_state.clone(),
basebackup_prepare_sender: self.basebackup_prepare_sender.clone(),
feature_resolver: self.feature_resolver.clone(),
}
}
@@ -5675,14 +5560,6 @@ impl TenantShard {
}
}
// Update metrics
let tid = self.tenant_shard_id.to_string();
let shard_id = self.tenant_shard_id.shard_slug().to_string();
let set_key = &[tid.as_str(), shard_id.as_str()][..];
TENANT_OFFLOADED_TIMELINES
.with_label_values(set_key)
.set(manifest.offloaded_timelines.len() as u64);
// Upload the manifest. Remote storage does no retries internally, so retry here.
match backoff::retry(
|| async {
@@ -5832,7 +5709,6 @@ pub(crate) mod harness {
pub conf: &'static PageServerConf,
pub tenant_conf: pageserver_api::models::TenantConfig,
pub tenant_shard_id: TenantShardId,
pub shard_identity: ShardIdentity,
pub generation: Generation,
pub shard: ShardIndex,
pub remote_storage: GenericRemoteStorage,
@@ -5900,7 +5776,6 @@ pub(crate) mod harness {
conf,
tenant_conf,
tenant_shard_id,
shard_identity,
generation,
shard,
remote_storage,
@@ -5951,8 +5826,6 @@ pub(crate) mod harness {
) -> anyhow::Result<Arc<TenantShard>> {
let walredo_mgr = Arc::new(WalRedoManager::from(TestRedoManager));
let (basebackup_requst_sender, _) = tokio::sync::mpsc::unbounded_channel();
let tenant = Arc::new(TenantShard::new(
TenantState::Attaching,
self.conf,
@@ -5962,15 +5835,14 @@ pub(crate) mod harness {
&ShardParameters::default(),
))
.unwrap(),
self.shard_identity,
// This is a legacy/test code path: sharding isn't supported here.
ShardIdentity::unsharded(),
Some(walredo_mgr),
self.tenant_shard_id,
self.remote_storage.clone(),
self.deletion_queue.new_client(),
// TODO: ideally we should run all unit tests with both configs
L0FlushGlobalState::new(L0FlushConfig::default()),
basebackup_requst_sender,
FeatureResolver::new_disabled(),
));
let preload = tenant
@@ -6084,7 +5956,6 @@ mod tests {
use timeline::compaction::{KeyHistoryRetention, KeyLogAtLsn};
use timeline::{CompactOptions, DeltaLayerTestDesc, VersionedKeySpaceQuery};
use utils::id::TenantId;
use utils::shard::{ShardCount, ShardNumber};
use super::*;
use crate::DEFAULT_PG_VERSION;
@@ -8413,24 +8284,10 @@ mod tests {
}
tline.freeze_and_flush().await?;
// Force layers to L1
tline
.compact(
&cancel,
{
let mut flags = EnumSet::new();
flags.insert(CompactFlags::ForceL0Compaction);
flags
},
&ctx,
)
.await?;
if iter % 5 == 0 {
let scan_lsn = Lsn(lsn.0 + 1);
info!("scanning at {}", scan_lsn);
let (_, before_delta_file_accessed) =
scan_with_statistics(&tline, &keyspace, scan_lsn, &ctx, io_concurrency.clone())
scan_with_statistics(&tline, &keyspace, lsn, &ctx, io_concurrency.clone())
.await?;
tline
.compact(
@@ -8439,14 +8296,13 @@ mod tests {
let mut flags = EnumSet::new();
flags.insert(CompactFlags::ForceImageLayerCreation);
flags.insert(CompactFlags::ForceRepartition);
flags.insert(CompactFlags::ForceL0Compaction);
flags
},
&ctx,
)
.await?;
let (_, after_delta_file_accessed) =
scan_with_statistics(&tline, &keyspace, scan_lsn, &ctx, io_concurrency.clone())
scan_with_statistics(&tline, &keyspace, lsn, &ctx, io_concurrency.clone())
.await?;
assert!(
after_delta_file_accessed < before_delta_file_accessed,
@@ -8740,10 +8596,8 @@ mod tests {
lsn: Lsn,
ctx: &RequestContext,
) -> Result<Option<Bytes>, GetVectoredError> {
let io_concurrency = IoConcurrency::spawn_from_conf(
tline.conf.get_vectored_concurrent_io,
tline.gate.enter().unwrap(),
);
let io_concurrency =
IoConcurrency::spawn_from_conf(tline.conf, tline.gate.enter().unwrap());
let mut reconstruct_state = ValuesReconstructState::new(io_concurrency);
let query = VersionedKeySpaceQuery::uniform(KeySpace::single(key..key.next()), lsn);
let mut res = tline
@@ -8887,8 +8741,6 @@ mod tests {
let cancel = CancellationToken::new();
// Image layer creation happens on the disk_consistent_lsn so we need to force set it now.
tline.force_set_disk_consistent_lsn(Lsn(0x40));
tline
.compact(
&cancel,
@@ -8902,7 +8754,8 @@ mod tests {
)
.await
.unwrap();
// Image layers are created at repartition LSN
// Image layers are created at last_record_lsn
let images = tline
.inspect_image_layers(Lsn(0x40), &ctx, io_concurrency.clone())
.await
@@ -9082,7 +8935,7 @@ mod tests {
.await;
// Update GC info
let mut guard = tline.gc_info.write().unwrap();
guard.cutoffs.time = Some(Lsn(0x30));
guard.cutoffs.time = Lsn(0x30);
guard.cutoffs.space = Lsn(0x30);
}
@@ -9190,7 +9043,7 @@ mod tests {
.await;
// Update GC info
let mut guard = tline.gc_info.write().unwrap();
guard.cutoffs.time = Some(Lsn(0x40));
guard.cutoffs.time = Lsn(0x40);
guard.cutoffs.space = Lsn(0x40);
}
tline
@@ -9420,77 +9273,6 @@ mod tests {
Ok(())
}
#[tokio::test]
async fn test_failed_flush_should_not_update_disk_consistent_lsn() -> anyhow::Result<()> {
//
// Setup
//
let harness = TenantHarness::create_custom(
"test_failed_flush_should_not_upload_disk_consistent_lsn",
pageserver_api::models::TenantConfig::default(),
TenantId::generate(),
ShardIdentity::new(ShardNumber(0), ShardCount(4), ShardStripeSize(128)).unwrap(),
Generation::new(1),
)
.await?;
let (tenant, ctx) = harness.load().await;
let timeline = tenant
.create_test_timeline(TIMELINE_ID, Lsn(0x10), DEFAULT_PG_VERSION, &ctx)
.await?;
assert_eq!(timeline.get_shard_identity().count, ShardCount(4));
let mut writer = timeline.writer().await;
writer
.put(
*TEST_KEY,
Lsn(0x20),
&Value::Image(test_img("foo at 0x20")),
&ctx,
)
.await?;
writer.finish_write(Lsn(0x20));
drop(writer);
timeline.freeze_and_flush().await.unwrap();
timeline.remote_client.wait_completion().await.unwrap();
let disk_consistent_lsn = timeline.get_disk_consistent_lsn();
let remote_consistent_lsn = timeline.get_remote_consistent_lsn_projected();
assert_eq!(Some(disk_consistent_lsn), remote_consistent_lsn);
//
// Test
//
let mut writer = timeline.writer().await;
writer
.put(
*TEST_KEY,
Lsn(0x30),
&Value::Image(test_img("foo at 0x30")),
&ctx,
)
.await?;
writer.finish_write(Lsn(0x30));
drop(writer);
fail::cfg(
"flush-layer-before-update-remote-consistent-lsn",
"return()",
)
.unwrap();
let flush_res = timeline.freeze_and_flush().await;
// if flush failed, the disk/remote consistent LSN should not be updated
assert!(flush_res.is_err());
assert_eq!(disk_consistent_lsn, timeline.get_disk_consistent_lsn());
assert_eq!(
remote_consistent_lsn,
timeline.get_remote_consistent_lsn_projected()
);
Ok(())
}
#[cfg(feature = "testing")]
#[tokio::test]
async fn test_simple_bottom_most_compaction_deltas_1() -> anyhow::Result<()> {
@@ -9679,7 +9461,7 @@ mod tests {
*guard = GcInfo {
retain_lsns: vec![],
cutoffs: GcCutoffs {
time: Some(Lsn(0x30)),
time: Lsn(0x30),
space: Lsn(0x30),
},
leases: Default::default(),
@@ -9763,7 +9545,7 @@ mod tests {
.await;
// Update GC info
let mut guard = tline.gc_info.write().unwrap();
guard.cutoffs.time = Some(Lsn(0x40));
guard.cutoffs.time = Lsn(0x40);
guard.cutoffs.space = Lsn(0x40);
}
tline
@@ -10234,7 +10016,7 @@ mod tests {
(Lsn(0x20), tline.timeline_id, MaybeOffloaded::No),
],
cutoffs: GcCutoffs {
time: Some(Lsn(0x30)),
time: Lsn(0x30),
space: Lsn(0x30),
},
leases: Default::default(),
@@ -10297,7 +10079,7 @@ mod tests {
let verify_result = || async {
let gc_horizon = {
let gc_info = tline.gc_info.read().unwrap();
gc_info.cutoffs.time.unwrap_or_default()
gc_info.cutoffs.time
};
for idx in 0..10 {
assert_eq!(
@@ -10375,7 +10157,7 @@ mod tests {
.await;
// Update GC info
let mut guard = tline.gc_info.write().unwrap();
guard.cutoffs.time = Some(Lsn(0x38));
guard.cutoffs.time = Lsn(0x38);
guard.cutoffs.space = Lsn(0x38);
}
tline
@@ -10483,7 +10265,7 @@ mod tests {
(Lsn(0x20), tline.timeline_id, MaybeOffloaded::No),
],
cutoffs: GcCutoffs {
time: Some(Lsn(0x30)),
time: Lsn(0x30),
space: Lsn(0x30),
},
leases: Default::default(),
@@ -10546,7 +10328,7 @@ mod tests {
let verify_result = || async {
let gc_horizon = {
let gc_info = tline.gc_info.read().unwrap();
gc_info.cutoffs.time.unwrap_or_default()
gc_info.cutoffs.time
};
for idx in 0..10 {
assert_eq!(
@@ -10732,7 +10514,7 @@ mod tests {
*guard = GcInfo {
retain_lsns: vec![(Lsn(0x18), branch_tline.timeline_id, MaybeOffloaded::No)],
cutoffs: GcCutoffs {
time: Some(Lsn(0x10)),
time: Lsn(0x10),
space: Lsn(0x10),
},
leases: Default::default(),
@@ -10752,7 +10534,7 @@ mod tests {
*guard = GcInfo {
retain_lsns: vec![(Lsn(0x40), branch_tline.timeline_id, MaybeOffloaded::No)],
cutoffs: GcCutoffs {
time: Some(Lsn(0x50)),
time: Lsn(0x50),
space: Lsn(0x50),
},
leases: Default::default(),
@@ -11473,7 +11255,7 @@ mod tests {
*guard = GcInfo {
retain_lsns: vec![(Lsn(0x20), tline.timeline_id, MaybeOffloaded::No)],
cutoffs: GcCutoffs {
time: Some(Lsn(0x30)),
time: Lsn(0x30),
space: Lsn(0x30),
},
leases: Default::default(),
@@ -11862,7 +11644,7 @@ mod tests {
(Lsn(0x20), tline.timeline_id, MaybeOffloaded::No),
],
cutoffs: GcCutoffs {
time: Some(Lsn(0x30)),
time: Lsn(0x30),
space: Lsn(0x30),
},
leases: Default::default(),
@@ -11925,7 +11707,7 @@ mod tests {
let verify_result = || async {
let gc_horizon = {
let gc_info = tline.gc_info.read().unwrap();
gc_info.cutoffs.time.unwrap_or_default()
gc_info.cutoffs.time
};
for idx in 0..10 {
assert_eq!(
@@ -12114,7 +11896,7 @@ mod tests {
(Lsn(0x20), tline.timeline_id, MaybeOffloaded::No),
],
cutoffs: GcCutoffs {
time: Some(Lsn(0x30)),
time: Lsn(0x30),
space: Lsn(0x30),
},
leases: Default::default(),
@@ -12177,7 +11959,7 @@ mod tests {
let verify_result = || async {
let gc_horizon = {
let gc_info = tline.gc_info.read().unwrap();
gc_info.cutoffs.time.unwrap_or_default()
gc_info.cutoffs.time
};
for idx in 0..10 {
assert_eq!(
@@ -12440,7 +12222,7 @@ mod tests {
*guard = GcInfo {
retain_lsns: vec![],
cutoffs: GcCutoffs {
time: Some(Lsn(0x30)),
time: Lsn(0x30),
space: Lsn(0x30),
},
leases: Default::default(),

View File

@@ -1348,21 +1348,6 @@ impl RemoteTimelineClient {
Ok(())
}
pub(crate) fn schedule_unlinking_of_layers_from_index_part<I>(
self: &Arc<Self>,
names: I,
) -> Result<(), NotInitialized>
where
I: IntoIterator<Item = LayerName>,
{
let mut guard = self.upload_queue.lock().unwrap();
let upload_queue = guard.initialized_mut()?;
self.schedule_unlinking_of_layers_from_index_part0(upload_queue, names);
Ok(())
}
/// Update the remote index file, removing the to-be-deleted files from the index,
/// allowing scheduling of actual deletions later.
fn schedule_unlinking_of_layers_from_index_part0<I>(

View File

@@ -668,9 +668,7 @@ impl From<DownloadError> for UpdateError {
impl From<std::io::Error> for UpdateError {
fn from(value: std::io::Error) -> Self {
if let Some(nix::errno::Errno::ENOSPC) =
value.raw_os_error().map(nix::errno::Errno::from_raw)
{
if let Some(nix::errno::Errno::ENOSPC) = value.raw_os_error().map(nix::errno::from_i32) {
UpdateError::NoSpace
} else if value
.get_ref()

View File

@@ -235,7 +235,7 @@ pub(super) async fn gather_inputs(
// than our internal space cutoff. This means that if someone drops a database and waits for their
// PITR interval, they will see synthetic size decrease, even if we are still storing data inside
// the space cutoff.
let mut next_pitr_cutoff = gc_info.cutoffs.time.unwrap_or_default(); // TODO: handle None
let mut next_pitr_cutoff = gc_info.cutoffs.time;
// If the caller provided a shorter retention period, use that instead of the GC cutoff.
let retention_param_cutoff = if let Some(max_retention_period) = max_retention_period {

View File

@@ -31,7 +31,6 @@ pub use inmemory_layer::InMemoryLayer;
pub(crate) use layer::{EvictionError, Layer, ResidentLayer};
pub use layer_desc::{PersistentLayerDesc, PersistentLayerKey};
pub use layer_name::{DeltaLayerName, ImageLayerName, LayerName};
use pageserver_api::config::GetVectoredConcurrentIo;
use pageserver_api::key::Key;
use pageserver_api::keyspace::{KeySpace, KeySpaceRandomAccum};
use pageserver_api::record::NeonWalRecord;
@@ -44,6 +43,7 @@ use self::inmemory_layer::InMemoryLayerFileId;
use super::PageReconstructError;
use super::layer_map::InMemoryLayerDesc;
use super::timeline::{GetVectoredError, ReadPath};
use crate::config::PageServerConf;
use crate::context::{
AccessStatsBehavior, PerfInstrumentFutureExt, RequestContext, RequestContextBuilder,
};
@@ -318,10 +318,11 @@ impl IoConcurrency {
}
pub(crate) fn spawn_from_conf(
conf: GetVectoredConcurrentIo,
conf: &'static PageServerConf,
gate_guard: GateGuard,
) -> IoConcurrency {
let selected = match conf {
use pageserver_api::config::GetVectoredConcurrentIo;
let selected = match conf.get_vectored_concurrent_io {
GetVectoredConcurrentIo::Sequential => SelectedIoConcurrency::Sequential,
GetVectoredConcurrentIo::SidecarTask => SelectedIoConcurrency::SidecarTask(gate_guard),
};

Some files were not shown because too many files have changed in this diff Show More