This test had two flaky failure modes:
- pageserver log error for timeline not found: this resulted from
changes for DR when timeline destroy/create was added, but endpoint was
left running during that operation.
- storage controller log error because the test was running for long
enough that a background reconcile happened at almost the exact moment
of test teardown, and our test fixtures tear down the pageservers before
the controller.
Closes: #7224
## Problem
Large quantities of ephemeral layer data can lead to excessive memory
consumption (https://github.com/neondatabase/neon/issues/6939). We
currently don't have a way to know how much ephemeral layer data is
present on a pageserver.
Before we can add new behaviors to proactively roll layers in response
to too much ephemeral data, we must calculate that total.
Related: https://github.com/neondatabase/neon/issues/6916
## Summary of changes
- Create GlobalResources and GlobalResourceUnits types, where timelines
carry a GlobalResourceUnits in their TimelineWriterState.
- Periodically update the size in GlobalResourceUnits:
- During tick()
- During layer roll
- During put() if the latest value has drifted more than 10MB since our
last update
- Expose the value of the global ephemeral layer bytes counter as a
prometheus metric.
- Extend the lifetime of TimelineWriterState:
- Instead of dropping it in TimelineWriter::drop, let it remain.
- Drop TimelineWriterState in roll_layer: this drops our guard on the
global byte count to reflect the fact that we're freezing the layer.
- Ensure the validity of the later in the writer state by clearing the
state in the same place we freeze layers, and asserting on the
write-ability of the layer in `writer()`
- Add a 'context' parameter to `get_open_layer_action` so that it can
skip the prev_lsn==lsn check when called in tick() -- this is needed
because now tick is called with a populated state, where
prev_lsn==Some(lsn) is true for an idle timeline.
- Extend layer rolling test to use this metric
- Remove code for using AWS secrets manager, as we're deploying with
k8s->env vars instead
- Load each secret independently, so that one can mix CLI args with
environment variables, rather than requiring that all secrets are loaded
with the same mechanism.
- Add a 'strict mode', enabled by default, which will refuse to start if
secrets are not loaded. This avoids the risk of accidentially disabling
auth by omitting the public key, for example
## Problem
We recently introduced log file validation for the storage controller.
The heartbeater will WARN when it fails
for a node, hence the test fails.
Closes https://github.com/neondatabase/neon/issues/7159
## Summary of changes
* Warn only once for each set of heartbeat retries
* Allow list heartbeat warns
Before this PR, each core had 3 executor threads from 3 different
runtimes. With this PR, we just have one runtime, with one thread per
core. Switching to a single tokio runtime should reduce that effective
over-commit of CPU and in theory help with tail latencies -- iff all
tokio tasks are well-behaved and yield to the runtime regularly.
Are All Tasks Well-Behaved? Are We Ready?
-----------------------------------------
Sadly there doesn't seem to be good out-of-the box tokio tooling to
answer this question.
We *believe* all tasks are well behaved in today's code base, as of the
switch to `virtual_file_io_engine = "tokio-epoll-uring"` in production
(https://github.com/neondatabase/aws/pull/1121).
The only remaining executor-thread-blocking code is walredo and some
filesystem namespace operations.
Filesystem namespace operations work is being tracked in #6663 and not
considered likely to actually block at this time.
Regarding walredo, it currently does a blocking `poll` for read/write to
the pipe file descriptors we use for IPC with the walredo process.
There is an ongoing experiment to make walredo async (#6628), but it
needs more time because there are surprisingly tricky trade-offs that
are articulated in that PR's description (which itself is still WIP).
What's relevant for *this* PR is that
1. walredo is always CPU-bound
2. production tail latencies for walredo request-response
(`pageserver_wal_redo_seconds_bucket`) are
- p90: with few exceptions, low hundreds of micro-seconds
- p95: except on very packed pageservers, below 1ms
- p99: all below 50ms, vast majority below 1ms
- p99.9: almost all around 50ms, rarely at >= 70ms
- [Dashboard
Link](https://neonprod.grafana.net/d/edgggcrmki3uof/2024-03-walredo-latency?orgId=1&var-ds=ZNX49CDVz&var-pXX_by_instance=0.9&var-pXX_by_instance=0.99&var-pXX_by_instance=0.95&var-adhoc=instance%7C%21%3D%7Cpageserver-30.us-west-2.aws.neon.tech&var-per_instance_pXX_max_seconds=0.0005&from=1711049688777&to=1711136088777)
The ones below 1ms are below our current threshold for when we start
thinking about yielding to the executor.
The tens of milliseconds stalls aren't great, but, not least because of
the implicit overcommit of CPU by the three runtimes, we can't be sure
whether these tens of milliseconds are inherently necessary to do the
walredo work or whether we could be faster if there was less contention
for CPU.
On the first item (walredo being always CPU-bound work): it means that
walredo processes will always compete with the executor threads.
We could yield, using async walredo, but then we hit the trade-offs
explained in that PR.
tl;dr: the risk of stalling executor threads through blocking walredo
seems low, and switching to one runtime cleans up one potential source
for higher-than-necessary stall times (explained in the previous
paragraphs).
Code Changes
------------
- Remove the 3 different runtime definitions.
- Add a new definition called `THE_RUNTIME`.
- Use it in all places that previously used one of the 3 removed
runtimes.
- Remove the argument from `task_mgr`.
- Fix failpoint usage where `pausable_failpoint!` should have been used.
We encountered some actual failures because of this, e.g., hung
`get_metric()` calls during test teardown that would client-timeout
after 300s.
As indicated by the comment above `THE_RUNTIME`, we could take this
clean-up further.
But before we create so much churn, let's first validate that there's no
perf regression.
Performance
-----------
We will test this in staging using the various nightly benchmark runs.
However, the worst-case impact of this change is likely compaction
(=>image layer creation) competing with compute requests.
Image layer creation work can't be easily generated & repeated quickly
by pagebench.
So, we'll simply watch getpage & basebackup tail latencies in staging.
Additionally, I have done manual benchmarking using pagebench.
Report:
https://neondatabase.notion.site/2024-03-23-oneruntime-change-benchmarking-22a399c411e24399a73311115fb703ec?pvs=4
Tail latencies and throughput are marginally better (no regression =
good).
Except in a workload with 128 clients against one tenant.
There, the p99.9 and p99.99 getpage latency is about 2x worse (at
slightly lower throughput).
A dip in throughput every 20s (compaction_period_ is clearly visible,
and probably responsible for that worse tail latency.
This has potential to improve with async walredo, and is an edge case
workload anyway.
Future Work
-----------
1. Once this change has shown satisfying results in production, change
the codebase to use the ambient runtime instead of explicitly
referencing `THE_RUNTIME`.
2. Have a mode where we run with a single-threaded runtime, so we
uncover executor stalls more quickly.
3. Switch or write our own failpoints library that is async-native:
https://github.com/neondatabase/neon/issues/7216
## Problem
The service that receives consumption metrics has lower availability
than S3. Writing metrics to S3 improves their availability.
Closes: https://github.com/neondatabase/cloud/issues/9824
## Summary of changes
- The same data as consumption metrics POST bodies is also compressed
and written to an S3 object with a timestamp-formatted path.
- Set `metric_collection_bucket` (same format as `remote_storage`
config) to configure the location to write to
A test was added which exercises secondary locations more, and there was
a location in the secondary downloader that warned on ephemeral files.
This was intended to be fixed in this faulty commit:
8cea866adf
This change improves the resilience of the system to unclean restarts.
Previously, re-attach responses only included attached tenants
- If the pageserver had local state for a secondary location, it would
remain, but with no guarantee that it was still _meant_ to be there.
After this change, the pageserver will only retain secondary locations
if the /re-attach response indicates that they should still be there.
- If the pageserver had local state for an attached location that was
omitted from a re-attach response, it would be entirely detached. This
is wasteful in a typical HA setup, where an offline node's tenants might
have been re-attached elsewhere before it restarts, but the offline
node's location should revert to a secondary location rather than being
wiped. Including secondary tenants in the re-attach response enables the
pageserver to avoid throwing away local state unnecessarily.
In this PR:
- The re-attach items are extended with a 'mode' field.
- Storage controller populates 'mode'
- Pageserver interprets it (default is attached if missing) to construct
either a SecondaryTenant or a Tenant.
- A new test exercises both cases.
Stacks on:
- https://github.com/neondatabase/neon/pull/7165
Fixes while working on background optimization of scheduling after a
split:
- When a tenant has secondary locations, we weren't detaching the parent
shards' secondary locations when doing a split
- When a reconciler detaches a location, it was feeding back a
locationconf with `Detached` mode in its `observed` object, whereas it
should omit that location. This could cause the background reconcile
task to keep kicking off no-op reconcilers forever (harmless but
annoying).
- During shard split, we were scheduling secondary locations for the
child shards, but no reconcile was run for these until the next time the
background reconcile task ran. Creating these ASAP is useful, because
they'll be used shortly after a shard split as the destination locations
for migrating the new shards to different nodes.
## Problem
Storage controller had basically no metrics.
## Summary of changes
1. Migrate the existing metrics to use Conrad's
[`measured`](https://docs.rs/measured/0.0.14/measured/) crate.
2. Add metrics for incoming http requests
3. Add metrics for outgoing http requests to the pageserver
4. Add metrics for outgoing pass through requests to the pageserver
5. Add metrics for database queries
Note that the metrics response for the attachment service does not use
chunked encoding like the rest of the metrics endpoints. Conrad has
kindly extended the crate such that it can now be done. Let's leave it
for a follow-up since the payload shouldn't be that big at this point.
Fixes https://github.com/neondatabase/neon/issues/6875
- Enable debug logs for this test
- Add some debug logging detail in downloader.rs
- Add an info-level message in scheduler.rs that makes it obvious if a
command is waiting for an existing task rather than spawning a new one.
Warm-up (and the "tenant startup complete" metric update) happens in
a background tokio task. The tenant map is eagerly updated (can happen
before the task finishes).
The test assumed that if the tenant map was updated, then the metric
should reflect that. That's not the case, so we tweak the test to wait
for the metric.
Fixes https://github.com/neondatabase/neon/issues/7158
This is a mixed bag of changes split out for separate review while
working on other things, and batched together to reduce load on CI
runners. Each commits stands alone for review purposes:
- do_tenant_shard_split was a long function and had a synchronous
validation phase at the start that could readily be pulled out into a
separate function. This also avoids the special casing of
ApiError::BadRequest when deciding whether an abort is needed on errors
- Add a 'describe' API (GET on tenant ID) that will enable storcon-cli
to see what's going on with a tenant
- the 'locate' API wasn't really meant for use in the field. It's for
tests: demote it to the /debug/ prefix
- The `Single` placement policy was a redundant duplicate of Double(0),
and Double was a bad name. Rename it Attached.
(https://github.com/neondatabase/neon/issues/7107)
- Some neon_local commands were added for debug/demos, which are now
replaced by commands in storcon-cli (#7114 ). Even though that's not
merged yet, we don't need the neon_local ones any more.
Closes https://github.com/neondatabase/neon/issues/7107
## Backward compat of Single/Double -> `Attached(n)` change
A database migration is used to convert any existing values.
## Problem
As with the pageserver, we should fail tests that emit unexpected log
errors/warnings.
## Summary of changes
- Refactor existing log checks to be reusable
- Run log checks for attachment_service
- Add allow lists as needed.
Add shard_number to PageserverFeedback and parse it on the compute side.
When compute receives a new ps_feedback, it calculates min LSNs among
feedbacks from all shards, and uses those LSNs for backpressure.
Add `test_sharding_backpressure` to verify that backpressure slows down
compute to wait for the slowest shard.
## Problem
The existing secondary download API relied on the caller to wait as long
as it took to complete -- for large shards that could be a long time, so
typical clients that might have a baked-in ~30s timeout would have a
problem.
## Summary of changes
- Take a `wait_ms` query parameter to instruct the pageserver how long
to wait: if the download isn't complete in this duration, then 201 is
returned instead of 200.
- For both 200 and 201 responses, include response body describing
download progress, in terms of layers and bytes. This is sufficient for
the caller to track how much data is being transferred and log/present
that status.
- In storage controller live migrations, use this API to apply a much
longer outer timeout, with smaller individual per-request timeouts, and
log the progress of the downloads.
- Add a test that injects layer download delays to exercise the new
behavior
## Problem
Tenant deletion had a couple of TODOs where we weren't using proper
cancellation tokens that would have aborted the deletions during process
shutdown.
## Summary of changes
- Refactor enough that deletion/shutdown code has access to the
TenantManager's cancellation toke
- Use that cancellation token in tenant deletion instead of dummy
tokens.
fixes https://github.com/neondatabase/neon/issues/7116
Changes:
- refactor PageServerConfigBuilder: support not-set values
- implement runtime feature test
- use runtime feature test to determine `virtual_file_io_engine` if not
explicitly configured in the config
- log the effective engine at startup
- drive-by: improve assertion messages in `test_pageserver_init_node_id`
This needed a tiny bit of tokio-epoll-uring work, hence bumping it.
Changelog:
```
git log --no-decorate --oneline --reverse 868d2c42b5d54ca82fead6e8f2f233b69a540d3e..342ddd197a060a8354e8f11f4d12994419fff939
c7a74c6 Bump mio from 0.8.8 to 0.8.11
4df3466 Bump mio from 0.8.8 to 0.8.11 (#47)
342ddd1 lifecycle: expose `LaunchResult` enum (#49)
```
## Problem
See:
- https://github.com/neondatabase/neon/issues/6374
## Summary of changes
Whereas previously we calculated synthetic size from the gc_horizon or
the pitr_interval (whichever is the lower LSN), now we ignore gc_horizon
and exclusively start from the `pitr_interval`. This is a more generous
calculation for billing, where we do not charge users for data retained
due to gc_horizon.
Switched the order; doing https://github.com/neondatabase/neon/pull/6139
first then can remove uninit marker after.
## Problem
Previously, existence of a timeline directory was treated as evidence of
the timeline's logical existence. That is no longer the case since we
treat remote storage as the source of truth on each startup: we can
therefore do without this mark file.
The mark file had also been used as a pseudo-lock to guard against
concurrent creations of the same TimelineId -- now that persistence is
no longer required, this is a bit unwieldy.
In #6139 the `Tenant::timelines_creating` was added to protect against
concurrent creations on the same TimelineId, making the uninit mark file
entirely redundant.
## Summary of changes
- Code that writes & reads mark file is removed
- Some nearby `pub` definitions are amended to `pub(crate)`
- `test_duplicate_creation` is added to demonstrate that mutual
exclusion of creations still works.
## Problem
We have no regression tests for websocket flow
## Summary of changes
Add a hacky implementation of the postgres protocol over websockets just
to verify the protocol behaviour does not regress over time.
## Problem
If a pageserver was offline when the storage controller started, there
was no mechanism to update the
storage controller state when the pageserver becomes active.
## Summary of changes
* Add a heartbeater module. The heartbeater must be driven by an
external loop.
* Integrate the heartbeater into the service.
- Extend the types used by the service and scheduler to keep track of a
nodes' utilisation score.
- Add a background loop to drive the heartbeater and update the state
based on the deltas it generated
- Do an initial round of heartbeats at start-up
# Problem
While investigating #7124, I noticed that the benchmark was always using
the `DEFAULT_*` `virtual_file_io_engine` , i.e., `tokio-epoll-uring` as
of https://github.com/neondatabase/neon/pull/7077.
The fundamental problem is that the `control_plane` code has its own
view of `PageServerConfig`, which, I believe, will always be a subset of
the real pageserver's `pageserver/src/config.rs`.
For the `virtual_file_io_engine` and `get_vectored_impl` parametrization
of the test suite, we were constructing a dict on the Python side that
contained these parameters, then handed it to
`control_plane::PageServerConfig`'s derived `serde::Deserialize`.
The default in serde is to ignore unknown fields, so, the Deserialize
impl silently ignored the fields.
In consequence, the fields weren't propagated to the `pageserver --init`
call, and the tests ended up using the
`pageserver/src/config.rs::DEFAULT_` values for the respective options
all the time.
Tests that explicitly used overrides in `env.pageserver.start()` and
similar were not affected by this.
But, it means that all the test suite runs where with parametrization
didn't properly exercise the code path.
# Changes
- use `serde(deny_unknown_fields)` to expose the problem
- With this change, the Python tests that override
`virtual_file_io_engine` and
`get_vectored_impl` fail on `pageserver --init`, exposing the problem.
- use destructuring to uncover the issue in the future
- fix the issue by adding the missing fields to the `control_plane`
crate's `PageServerConf`
- A better solution would be for control plane to re-use a struct
provided
by the pageserver crate, so that everything is in one place in
`pageserver/src/config.rs`, but, our config parsing code is (almost)
beyond repair anyways.
- fix the `pageserver_virtual_file_io_engine` to be responsive to the
env var
- => required to make parametrization work in benchmarks
# Testing
Before merging this PR, I re-ran the regression tests & CI with the full
matrix of `virtual_file_io_engine` and `tokio-epoll-uring`, see
9c7ea364e0
## Problem
Shard splits worked, but weren't safe against failures (e.g. node crash
during split) yet.
Related: #6676
## Summary of changes
- Introduce async rwlocks at the scope of Tenant and Node:
- exclusive tenant lock is used to protect splits
- exclusive node lock is used to protect new reconciliation process that
happens when setting node active
- exclusive locks used in both cases when doing persistent updates (e.g.
node scheduling conf) where the update to DB & in-memory state needs to
be atomic.
- Add failpoints to shard splitting in control plane and pageserver
code.
- Implement error handling in control plane for shard splits: this
detaches child chards and ensures parent shards are re-attached.
- Crash-safety for storage controller restarts requires little effort:
we already reconcile with nodes over a storage controller restart, so as
long as we reset any incomplete splits in the DB on restart (added in
this PR), things are implicitly cleaned up.
- Implement reconciliation with offline nodes before they transition to
active:
- (in this context reconciliation means something like
startup_reconcile, not literally the Reconciler)
- This covers cases where split abort cannot reach a node to clean it
up: the cleanup will eventually happen when the node is marked active,
as part of reconciliation.
- This also covers the case where a node was unavailable when the
storage controller started, but becomes available later: previously this
allowed it to skip the startup reconcile.
- Storage controller now terminates on panics. We only use panics for
true "should never happen" assertions, and these cases can leave us in
an un-usable state if we keep running (e.g. panicking in a shard split).
In the unlikely event that we get into a crashloop as a result, we'll
rely on kubernetes to back us off.
- Add `test_sharding_split_failures` which exercises a variety of
failure cases during shard split.
## Problem
hyper auto-cancels the request futures on connection close.
`sql_over_http::handle` is not 'drop cancel safe', so we need to do some
other work to make sure connections are queries in the right way.
## Summary of changes
1. tokio::spawn the request handler to resolve the initial cancel-safety
issue
2. share a cancellation token, and cancel it when the request `Service`
is dropped.
3. Add a new log span to be able to track the HTTP connection lifecycle.
## Summary
- Currently we can set stripe size at tenant creation, but it doesn't
mean anything until we have multiple shards
- When onboarding an existing tenant, it will always get a default shard
stripe size, so we would like to be able to pick the actual stripe size
at the point we split.
## Why do this inline with a split?
The alternative to this change would be to have a separate endpoint on
the storage controller for setting the stripe size on a tenant, and only
permit writes to that endpoint when the tenant has only a single shard.
That would work, but be a little bit more work for a client, and not
appreciably simpler (instead of having a special argument to the split
functions, we'd have a special separate endpoint, and a requirement that
the controller must sync its config down to the pageserver before
calling the split API). Either approach would work, but this one feels a
bit more robust end-to-end: the split API is the _very last moment_ that
the stripe size is mutable, so if we aim to set it before splitting, it
makes sense to do it as part of the same operation.
## Problem
Currently we manually register nodes with the storage controller, and
use a script during deploy to register with the cloud control plane.
Rather than extend that script further, nodes should just register on
startup.
## Summary of changes
- Extend the re-attach request to include an optional
NodeRegisterRequest
- If the `register` field is set, handle it like a normal node
registration before executing the normal re-attach work.
- Update tests/neon_local that used to rely on doing an explicit
register step that could be enabled/disabled.
---------
Co-authored-by: Christian Schwarz <christian@neon.tech>
## Problem
On HTTP query timeout, we should try and cancel the current in-flight
SQL query.
## Summary of changes
Trigger a cancellation command in postgres once the timeout is reach
Not a user-facing change, but can break any existing `.neon` directories
created by neon_local, as the name of the database used by the storage
controller changes.
This PR changes all the locations apart from the path of
`control_plane/attachment_service` (waiting for an opportune moment to
do that one, because it's the most conflict-ish wrt ongoing PRs like
#6676 )
This test occasionally fails with a difference in "pg_xact/0000" file
between the local and restored datadirs. My hypothesis is that
something changed in the database between the last explicit checkpoint
and the shutdown. I suspect autovacuum, it could certainly create
transactions.
To fix, be more precise about the point in time that we compare. Shut
down the endpoint first, then read the last LSN (i.e. the shutdown
checkpoint's LSN), from the local disk with pg_controldata. And use
exactly that LSN in the basebackup.
Closes#559
The walproposer pretends to be a walsender in many ways. It has a
WalSnd slot, it claims to be a walsender by calling
MarkPostmasterChildWalSender() etc. But one different to real
walsenders was that the postmaster still treated it as a bgworker
rather than a walsender. The difference is that at shutdown,
walsenders are not killed until the very end, after the checkpointer
process has written the shutdown checkpoint and exited.
As a result, the walproposer always got killed before the shutdown
checkpoint was written, so the shutdown checkpoint never made it to
safekeepers. That's fine in principle, we don't require a clean
shutdown after all. But it also feels a bit silly not to stream the
shutdown checkpoint. It could be useful for initializing hot standby
mode in a read replica, for example.
Change postmaster to treat background workers that have called
MarkPostmasterChildWalSender() as walsenders. That unfortunately
requires another small change in postgres core.
After doing that, walproposers stay alive longer. However, it also
means that the checkpointer will wait for the walproposer to switch to
WALSNDSTATE_STOPPING state, when the checkpointer sends the
PROCSIG_WALSND_INIT_STOPPING signal. We don't have the machinery in
walproposer to receive and handle that signal reliably. Instead, we
mark walproposer as being in WALSNDSTATE_STOPPING always.
In commit 568f91420a, I assumed that shutdown will wait for all the
remaining WAL to be streamed to safekeepers, but before this commit
that was not true, and the test became flaky. This should make it
stable again.
Some tests wrongly assumed that no WAL could have been written between
pg_current_wal_flush_lsn and quick pg stop after it. Fix them by introducing
flush_ep_to_pageserver which first stops the endpoint and then waits till all
committed WAL reaches the pageserver.
In passing extract safekeeper http client to its own module.
All of production is using it now as of
https://github.com/neondatabase/aws/pull/1121
The change in `flaky_tests.py` resets the flakiness detection logic.
The alternative would have been to repeat the choice of io engine in
each test name, which would junk up the various test reports too much.
---------
Co-authored-by: Alexander Bayandin <alexander@neon.tech>
## Problem
Before this PR, it was possible that on-demand downloads were started
after `Timeline::shutdown()`.
For example, we have observed a walreceiver-connection-handler-initiated
on-demand download that was started after `Timeline::shutdown()`s final
`task_mgr::shutdown_tasks()` call.
The underlying issue is that `task_mgr::shutdown_tasks()` isn't sticky,
i.e., new tasks can be spawned during or after
`task_mgr::shutdown_tasks()`.
Cc: https://github.com/neondatabase/neon/issues/4175 in lieu of a more
specific issue for task_mgr. We already decided we want to get rid of it
anyways.
Original investigation:
https://neondb.slack.com/archives/C033RQ5SPDH/p1709824952465949
## Changes
- enter gate while downloading
- use timeline cancellation token for cancelling download
thereby, fixes#7054
Entering the gate might also remove recent "kept the gate from closing"
in staging.
## Problem
When we start compute with newer version of extension (i.e. 1.2) and
then rollback the release, downgrading the compute version, next compute
start will try to update extension to the latest version available in
neon.control (i.e. 1.1).
Thus we need to provide downgrade scripts like neon--1.2--1.1.sql
These scripts must revert the changes made by the upgrade scripts in the
reverse order. This is necessary to ensure that the next upgrade will
work correctly.
In general, we need to write upgrade and downgrade scripts to be more
robust and add IF EXISTS / CREATE OR REPLACE clauses to all statements
(where applicable).
## Summary of changes
Adds downgrade scripts.
Adds test cases for extension downgrade/upgrade.
fixes#7066
This is a follow-up for
https://app.incident.io/neondb/incidents/167?tab=follow-ups
Signed-off-by: Alex Chi Z <chi@neon.tech>
Co-authored-by: Alex Chi Z <iskyzh@gmail.com>
Co-authored-by: Anastasia Lubennikova <anastasia@neon.tech>
## Problem
We reverted https://github.com/neondatabase/neon/pull/6661 a few days
ago. The change led to OOMs in
benchmarks followed by large WAL reingests.
The issue was that we removed [this
code](d04af08567/pageserver/src/tenant/timeline/walreceiver/walreceiver_connection.rs (L409-L417)).
That call may trigger a roll of the open layer due to
the keepalive messages received from the safekeeper. Removing it meant
that enforcing
of checkpoint timeout became even more lax and led to using up large
amounts of memory
for the in memory layer indices.
## Summary of changes
Piggyback on keep alive messages to enforce checkpoint timeout. This is
a hack, but it's exactly what
the current code is doing.
## Alternatives
Christhian, Joonas and myself sketched out a timer based approach
[here](https://github.com/neondatabase/neon/pull/6940). While discussing
it further, it became obvious that's also a bit of a hack and not the
desired end state. I chose not
to take that further since it's not what we ultimately want and it'll be
harder to rip out.
Right now it's unclear what the ideal system behaviour is:
* early flushing on memory pressure, or ...
* detaching tenants on memory pressure
We have a benchmark for creating a lot of branches, but it does random
things, and the branch count is not what we is the largest maximum we
aim to support. If this PR would stabilize the benchmark total duration
it means that there are some structures which are very much slower than
others. Then we should add a seed-outputting variant to help find and
reproduce such cases.
Additionally, record for the benchmark:
- shutdown duration
- startup metrics once done (on restart)
- duration of first compaction completion via debug logging
## Problem
- The storage controller is the source of truth for a tenant's stripe
size, but doesn't currently have a way to propagate that to compute:
we're just using the default stripe size everywhere.
Closes: https://github.com/neondatabase/neon/issues/6903
## Summary of changes
- Include stripe size in `ComputeHookNotifyRequest`
- Include stripe size in `LocationConfigResponse`
The stripe size is optional: it will only be advertised for
multi-sharded tenants. This enables the controller to defer the choice
of stripe size until we split a tenant for the first time.
part of https://github.com/neondatabase/neon/issues/5899
Problem
-------
Before this PR, the time spent waiting on the throttle was charged
towards the higher-level page_service metrics, i.e.,
`pageserver_smgr_query_seconds`.
The metrics are the foundation of internal SLIs / SLOs.
A throttled tenant would cause the SLI to degrade / SLO alerts to fire.
Changes
-------
- don't charge time spent in throttle towards the page_service metrics
- record time spent in throttle in RequestContext and subtract it from
the elapsed time
- this works because the page_service path doesn't create child context,
so, all the throttle time is recorded in the parent
- it's quite brittle and will break if we ever decide to spawn child
tasks that need child RequestContexts, which would have separate
instances of the `micros_spent_throttled` counter.
- however, let's punt that to a more general refactoring of
RequestContext
- add a test case that ensures that
- throttling happens for getpage requests; this aspect of the test
passed before this PR
- throttling delays aren't charged towards the page_service metrics;
this aspect of the test only passes with this PR
- drive-by: make the throttle log message `info!`, it's an expected
condition
Performance
-----------
I took the same measurements as in #6706 , no meaningful change in CPU
overhead.
Future Work
-----------
This PR enables us to experiment with the throttle for select tenants
without affecting the SLI metrics / triggering SLO alerts.
Before declaring this feature done, we need more work to happen,
specifically:
- decide on whether we want to retain the flexibility of throttling any
`Timeline::get` call, filtered by TaskKind
- versus: separate throttles for each page_service endpoint, potentially
with separate config options
- the trouble here is that this decision implies changes to the
TenantConfig, so, if we start using the current config style now, then
decide to switch to a different config, it'll be a breaking change
Nice-to-haves but probably not worth the time right now:
- Equivalent tests to ensure the throttle applies to all other
page_service handlers.
## Problem
- #6966
- Existing logs aren't pointing to a cause: it looks like heatmap upload
and download are happening, but for some reason the evicted layer isn't
removed on the secondary location.
## Summary of changes
- Assert evicted layer is gone from heatmap before checking its gone
from local disk: this will give clarity on whether the issue is with the
uploads or downloads.
- On assertion failures, log the contents of heatmap.
During onboarding, the control plane may attempt ad-hoc creation of a
secondary location to facilitate live migration. This gives us two
problems to solve:
- Accept 'Secondary' mode in /location_config and use it to put the
tenant into secondary mode on some physical pageserver, then pass
through /tenant/xyz/secondary/download requests
- Create tenants with no generation initially, since the initial
`Secondary` mode call will not provide us a generation.
This PR also fixes modification of a tenant's TenantConf during
/location_conf, which was previously ignored, and refines the flow for
config modification:
- avoid bumping generations when the only reason we're reconciling an
attached location is a config change
- increment TenantState.sequence when spawning a reconciler: usually
schedule() does this, but when we do config changes that doesn't happen,
so without this change waiters would think reconciliation was done
immediately. `sequence` is a bit of a murky thing right now, as it's
dual-purposed for tracking waiters, and for checking if an existing
reconciliation is already making updates to our current sequence. I'll
follow up at some point to clarify it's purpose.
- test config modification at the end of onboarding test
