## Problem
As reported in #10441 the `control_plane/README/md` incorrectly
specified that `--pg-version` should be specified in the `cargo neon
init` command. This is not the case and causes an invalid argument
error.
## Summary of changes
Fix the README
## Test Plan
I verified that the steps in the README now work locally. I connected to
the started postgres endpoint and executed some basic metadata queries.
## Problem
Close LKB-270. This is part of our series of efforts to make sure
lsn_lease API prompts clients to retry. Follow up of
https://github.com/neondatabase/neon/pull/12631.
Slack thread w/ Vlad:
https://databricks.slack.com/archives/C09254R641L/p1752677940697529
## Summary of changes
- Use `tenant_remote_mutation` API for LSN leases. Makes it consistent
with new APIs added to storcon.
- For 404, we now always retry because we know the tenant is
to-be-attached and will eventually reach a point that we can find that
tenant on the intent pageserver.
- Using the `tenant_remote_mutation` API also prevents us from the case
where the intent pageserver changes within the lease request. The
wrapper function will error with 503 if such things happen.
---------
Signed-off-by: Alex Chi Z <chi@neon.tech>
## Problem
A high rate of short-lived connections means that there a lot of cancel
keys in Redis with TTL=10min that could be avoided by having a much
shorter initial TTL.
## Summary of changes
* Introduce an initial TTL of 1min used with the SET command.
* Fix: don't delay repushing cancel data when expired.
* Prepare for exponentially increasing TTLs.
## Alternatives
A best-effort UNLINK command on connection termination would clean up
cancel keys right away. This needs a bigger refactor due to how batching
is handled.
## Problem
We currently offload LFC state unconditionally, which can cause
problems. Imagine a situation:
1. Endpoint started with `autoprewarm: true`.
2. While prewarming is not completed, we upload the new incomplete
state.
3. Compute gets interrupted and restarts.
4. We start again and try to prewarm with the state from 2. instead of
the previous complete state.
During the orchestrated prewarming, it's probably not a big issue, but
it's still better to do not interfere with the prewarm process.
## Summary of changes
Do not offload LFC state if we are currently prewarming or any issue
occurred. While on it, also introduce `Skipped` LFC prewarm status,
which is used when the corresponding LFC state is not present in the
endpoint storage. It's primarily needed to distinguish the first compute
start for particular endpoint, as it's completely valid to do not have
LFC state yet.
## Problem
Previously, if a get page failure was cause by timeline shutdown, the
pageserver would attempt to tear down the connection gracefully:
`shutdown(SHUT_WR)` followed by `close()`.
This triggers a code path on the compute where it has to tell apart
between an idle connection and a closed one. That code is bug prone, so
we can just side-step the issue by shutting down the connection via a
libpq error message.
This surfaced as instability in test_shard_resolve_during_split_abort.
It's a new test, but the issue existed for ages.
## Summary of Changes
Send a libpq error message instead of doing graceful TCP connection
shutdown.
Closes LKB-648
## Problem
See https://databricks.slack.com/archives/C09254R641L/p1752004515032899
stripe_size GUC update may be delayed at different backends and so cause
inconsistency with connection strings (shard map).
## Summary of changes
Postmaster should store stripe_size in shared memory as well as
connection strings.
It should be also enforced that stripe size is defined prior to
connection strings in postgresql.conf
---------
Co-authored-by: Konstantin Knizhnik <knizhnik@neon.tech>
Co-authored-by: Kosntantin Knizhnik <konstantin.knizhnik@databricks.com>
## Problem
Initializing of shared memory in extension is complex and non-portable.
In neon extension this boilerplate code is duplicated in several files.
## Summary of changes
Perform all initialization in one place - neon.c
All other module procvide *ShmemRequest() and *ShmemInit() fuinction
which are called from neon.c
---------
Co-authored-by: Kosntantin Knizhnik <konstantin.knizhnik@databricks.com>
Co-authored-by: Heikki Linnakangas <heikki@neon.tech>
## Problem
Follow up of https://github.com/neondatabase/neon/pull/12620
Discussions:
https://databricks.slack.com/archives/C09254R641L/p1752677940697529
The original code and after the patch above we converts 404s to 503s
regardless of the type of 404. We should only do that for tenant not
found errors. For other 404s like timeline not found, we should not
prompt clients to retry.
## Summary of changes
- Inspect the response body to figure out the type of 404. If it's a
tenant not found error, return 503.
- Otherwise, fallthrough and return 404 as-is.
- Add `tenant_shard_remote_mutation` that manipulates a single shard.
- Use `Service::tenant_shard_remote_mutation` for tenant shard
passthrough requests. This prevents us from another race that the attach
state changes within the request. (This patch mainly addresses the case
that the tenant is "not yet attached").
- TODO: lease API is still using the old code path. We should refactor
it to use `tenant_remote_mutation`.
---------
Signed-off-by: Alex Chi Z <chi@neon.tech>
Putting this in the neon codebase for now, to experiment. Can be lifted
into measured at a later date.
This metric family is like a MetricVec, but it only supports 1 label
being set at a time. It is useful for reporting info, rather than
reporting metrics.
https://www.robustperception.io/exposing-the-software-version-to-prometheus/
Second PR for the hashmap behind the updated LFC implementation ([see
first here](https://github.com/neondatabase/neon/pull/12595)). This only
adds the raw code for the hashmap/lock implementations and doesn't plug
it into the crate (that's dependent on the previous PR and should
probably be done when the full integration into the new communicator is
merged alongside `communicator-rewrite` changes?).
Some high level details: the communicator codebase expects to be able to
store references to entries within this hashmap for arbitrary periods of
time and so the hashmap cannot be allowed to move them during a rehash.
As a result, this implementation has a slightly unusual structure where
key-value pairs (and hash chains) are allocated in a separate region
with a freelist. The core hashmap structure is then an array of
"dictionary entries" that are just indexes into this region of key-value
pairs.
Concurrency support is very naive at the moment with the entire map
guarded by one big `RwLock` (which is implemented on top of a
`pthread_rwlock_t` since Rust doesn't guarantee that a
`std::sync::RwLock` is safe to use in shared memory). This (along with a
lot of other things) is being changed on the
`quantumish/lfc-resizable-map` branch.
## Problem
The `keep_failing_reconciles` counter was introduced in #12391, but
there is a special case:
> if a reconciliation loop claims to have succeeded, but maybe_reconcile
still thinks the tenant is in need of reconciliation, then that's a
probable bug and we should activate a similar backoff to prevent
flapping.
This PR redefines "flapping" to include not just repeated failures, but
also consecutive reconciliations of any kind (success or failure).
## Summary of Changes
- Replace `keep_failing_reconciles` with a new `stuck_reconciles` metric
- Replace `MAX_CONSECUTIVE_RECONCILIATION_ERRORS` with
`MAX_CONSECUTIVE_RECONCILES`, and increasing that from 5 to 10
- Increment the consecutive reconciles counter for all reconciles, not
just failures
- Reset the counter in `reconcile_all` when no reconcile is needed for a
shard
- Improve and fix the related test
---------
Co-authored-by: Aleksandr Sarantsev <aleksandr.sarantsev@databricks.com>
## Problem
The forward compatibility test is erroneously
using the downloaded (old) compatibility data. This test is meant to
test that old binaries can work with **new** data. Using the old
compatibility data renders this test useless.
## Summary of changes
Use new snapshot in test_forward_compat
Closes LKB-666
Co-authored-by: William Huang <william.huang@databricks.com>
## Problem
The force deletion API should behave like the graceful deletion API - it
needs to support cancellation, persistence, and be non-blocking.
## Summary of Changes
- Added a `force` flag to the `NodeStartDelete` command.
- Passed the `force` flag through the `start_node_delete` handler in the
storage controller.
- Handled the `force` flag in the `delete_node` function.
- Set the tombstone after removing the node from memory.
- Minor cleanup, like adding a `get_error_on_cancel` closure.
---------
Co-authored-by: Aleksandr Sarantsev <aleksandr.sarantsev@databricks.com>
## Problem
part of https://github.com/neondatabase/neon/issues/11318 ; it is not
entirely safe to run gc-compaction over the metadata key range due to
tombstones and implications of image layers (missing key in image layer
== key not exist). The auto gc-compaction trigger already skips metadata
key ranges (see `schedule_auto_compaction` call in
`trigger_auto_compaction`). In this patch we enforce it directly in
gc_compact_inner so that compactions triggered via HTTP API will also be
subject to this restriction.
## Summary of changes
Ensure gc-compaction only runs on rel key ranges.
Signed-off-by: Alex Chi Z <chi@neon.tech>
## Problem
`make neon-pgindent` doesn't work:
- there's no `$(BUILD_DIR)/neon-v17` dir
- `make -C ...` along with relative `BUILD_DIR` resolves to a path that
doesn't exist
## Summary of changes
- Fix path for to neon extension for `make neon-pgindent`
- Make `BUILD_DIR` absolute
- Remove trailing slash from `POSTGRES_INSTALL_DIR` to avoid duplicated
slashed in commands (doesn't break anything, it make it look nicer)
This PR simplifies our node info cache. Now we'll store entries for at
most the TTL duration, even if Redis notifications are available. This
will allow us to cache intermittent errors later (e.g. due to rate
limits) with more predictable behavior.
Related to https://github.com/neondatabase/cloud/issues/19353
## Problem
We were only resetting the limit in the wal proposer. If backends are
back pressured, it might take a while for the wal proposer to receive a
new WAL to reset the limit.
## Summary of changes
Backend also checks the time and resets the limit.
## How is this tested?
pgbench has more smooth tps
Signed-off-by: Tristan Partin <tristan.partin@databricks.com>
Co-authored-by: Haoyu Huang <haoyu.huang@databricks.com>
## Problem
close LKB-270, close LKB-253
We periodically saw pageserver returns 404 -> storcon converts it to 500
to cplane, and causing branch operations fail. This is due to storcon is
migrating tenants across pageservers and the request was forwarded from
the storcon to pageservers while the tenant was not attached yet. Such
operations should be retried from cplane and storcon should return 503
in such cases.
## Summary of changes
- Refactor `tenant_timeline_lsn_lease` to have a single function process
and passthrough such requests: `collect_tenant_shards` for collecting
all shards and checking if they're consistent with the observed state,
`process_result_and_passthrough_errors` to convert 404 into 503 if
necessary.
- `tenant_shard_node` also checks observed state now.
Note that for passthrough shard0, we originally had a check to convert
404 to 503:
```
// Transform 404 into 503 if we raced with a migration
if resp.status() == reqwest::StatusCode::NOT_FOUND {
// Look up node again: if we migrated it will be different
let new_node = service.tenant_shard_node(tenant_shard_id).await?;
if new_node.get_id() != node.get_id() {
// Rather than retry here, send the client a 503 to prompt a retry: this matches
// the pageserver's use of 503, and all clients calling this API should retry on 503.
return Err(ApiError::ResourceUnavailable(
format!("Pageserver {node} returned 404, was migrated to {new_node}").into(),
));
}
}
```
However, this only checks the intent state. It is possible that the
migration is in progress before/after the request is processed and
intent state is always the same throughout the API call, therefore 404
not being processed by this branch.
Also, not sure about if this new code is correct or not, need second
eyes on that:
```
// As a reconciliation is in flight, we do not have the observed state yet, and therefore we assume it is always inconsistent.
Ok((node.clone(), false))
```
---------
Signed-off-by: Alex Chi Z <chi@neon.tech>
## Problem
The warning message was seen during deployment, but it's actually OK.
## Summary of changes
- Treat `"No broker updates received for a while ..."` as an info
message.
Co-authored-by: Aleksandr Sarantsev <aleksandr.sarantsev@databricks.com>
N.B: No-op for the neon-env.
## Problem
We added a per-timeline disk utilization protection circuit breaker,
which will stop the safekeeper from accepting more WAL writes if the
disk utilization by the timeline has exceeded a configured limit. We
mainly designed the mechanism as a guard against WAL upload/backup bugs,
and we assumed that as long as WAL uploads are proceeding as normal we
will not run into disk pressure. This turned out to be not true. In one
of our load tests where we have 500 PGs ingesting data at the same time,
safekeeper disk utilization started to creep up even though WAL uploads
were completely normal (we likely just maxed out our S3 upload bandwidth
from the single SK). This means the per-timeline disk utilization
protection won't be enough if too many timelines are ingesting data at
the same time.
## Summary of changes
Added a global disk utilization protection circuit breaker which will
stop a safekeeper from accepting more WAL writes if the total disk usage
on the safekeeper (across all tenants) exceeds a limit. We implemented
this circuit breaker through two parts:
1. A "global disk usage watcher" background task that runs at a
configured interval (default every minute) to see how much disk space is
being used in the safekeeper's filesystem. This background task also
performs the check against the limit and publishes the result to a
global atomic boolean flag.
2. The `hadron_check_disk_usage()` routine (in `timeline.rs`) now also
checks this global boolean flag published in the step above, and fails
the `WalAcceptor` (triggers the circuit breaker) if the flag was raised.
The disk usage limit is disabled by default.
It can be tuned with the `--max-global-disk-usage-ratio` CLI arg.
## How is this tested?
Added integration test
`test_wal_acceptor.py::test_global_disk_usage_limit`.
Also noticed that I haven't been using the `wait_until(f)` test function
correctly (the `f` passed in is supposed to raise an exception if the
condition is not met, instead of returning `False`...). Fixed it in both
circuit breaker tests.
---------
Co-authored-by: William Huang <william.huang@databricks.com>
## Problem
In the gap between picking an optimization and applying it, something
might insert a change to the intent state that makes it incompatible.
If the change is done via the `schedule()` method, we are covered by the
increased sequence number, but otherwise we can panic if we violate the
intent state invariants.
## Summary of Changes
Validate the optimization right before applying it. Since we hold the
service lock at that point, nothing else can sneak in.
Closes LKB-65
## Problem
We run multiple proxies, we get logs like
```
... spans={"http_conn#22":{"conn_id": ...
... spans={"http_conn#24":{"conn_id": ...
```
these are the same span, and the difference is confusing.
## Summary of changes
Introduce a counter per span name, rather than a global counter. If the
counter is 0, no change to the span name is made.
To follow up: see which span names are duplicated within the codebase in
different callsites
## Problem
We have several linters that use Node.js, but they are currently set up
differently, both locally and on CI.
## Summary of changes
- Add Node.js to `build-tools` image
- Move `compute/package.json` -> `build-tools/package.json` and add
`redocly` to it `@redocly/cli`
- Unify and merge into one job `lint-openapi-spec` and
`validate-compute-manifest`
## Problem
neondatabase/neon#12601 did't compleatly disable writing `*.profraw`
files, but instead of `/tmp/coverage` it started to write into the
current directory
## Summary of changes
- Set `LLVM_PROFILE_FILE=/dev/null` to avoing writing `*.profraw` at all
Initial PR for the hashmap behind the updated LFC implementation. This
refactors `neon-shmem` so that the actual shared memory utilities are in
a separate module within the crate. Beyond that, it slightly changes
some of the docstrings so that they play nicer with `cargo doc`.
We don't want to depend on postgres_ffi in an API crate. If there is no
such dependency, we can compile stuff like `storcon_cli` without needing
a full working postgres build. Fixes regression of #12548 (before we
could compile it).
- Remove a few obsolete "allowed error messages" from tests. The
pageserver doesn't emit those messages anymore.
- Remove misplaced and outdated docstring comment from
`test_tenants.py`. A docstring is supposed to be the first thing in a
function, but we had added some code before it. And it was outdated, as
we haven't supported running without safekeepers for a long time.
- Fix misc typos in comments
- Remove obsolete comment about backwards compatibility with safekeepers
without `TIMELINE_STATUS` API. All safekeepers have it by now.
## Problem
We don't use code coverage produced by `regress-tests`
(neondatabase/neon#6798), so there's no need to collect it. Potentially,
disabling it should reduce the load on disks and improve the stability
of debug builds.
## Summary of changes
- Disable code coverage collection for regression tests
## Problem
This is a prerequisite for neondatabase/neon#12575 to keep all things
relevant to `build-tools` image in a single directory
## Summary of changes
- Rename `build_tools/` to `build-tools/`
- Move `build-tools.Dockerfile` to `build-tools/Dockerfile`
## Problem
Not all cplane errors are properly recognized and cached/retried.
## Summary of changes
Add more cplane error reasons. Also, use retry_delay_ms as cache TTL if
present.
Related to https://github.com/neondatabase/cloud/issues/19353
All Errors that can occur during get_installed_extensions() come from
tokio-postgres functions, e.g. if the database is being shut down
("FATAL: terminating connection due to administrator command"). I'm
seeing a lot of such errors in the logs with the regression tests, with
very verbose stack traces. The compute_ctl stack trace is pretty useless
for errors originating from the Postgres connection, the error message
has all the information, so stop printing the stack trace.
I changed the result type of the functions to return the originating
tokio_postgres Error rather than anyhow::Error, so that if we introduce
other error sources to the functions where the stack trace might be
useful, we'll be forced to revisit this, probably by introducing a new
Error type that separates postgres errors from other errors. But this
will do for now.
We didn't consistently apply these, and it wasn't consistently solved.
With this patch we should have a more consistent approach to this, and
have less issues porting changes to newer versions.
This also removes some potentially buggy casts to `long` from `uint64` -
they could've truncated the value in systems where `long` only has 32
bits.
Include records and image in the debug get page handler.
This endpoint does not update the metrics and does not support tracing.
Note that this now returns individual bytes which need to be encoded
properly for debugging.
Co-authored-by: Haoyu Huang <haoyu.huang@databricks.com>
# TLDR
This PR is a no-op.
## Problem
When a SK loses a disk, it must recover all WALs from the very
beginning. This may take days/weeks to catch up to the latest WALs for
all timelines it owns.
## Summary of changes
When SK starts up,
if it finds that it has 0 timelines,
- it will ask SC for the timeline it owns.
- Then, pulls the timeline from its peer safekeepers to restore the WAL
redundancy right away.
After pulling timeline is complete, it will become active and accepts
new WALs.
The current impl is a prototype. We can optimize the impl further, e.g.,
parallel pull timelines.
---------
Co-authored-by: Haoyu Huang <haoyu.huang@databricks.com>
https://github.com/neondatabase/cloud/issues/19011
Measure relative performance for prewarmed and non-prewarmed endpoints.
Add test that runs on every commit, and one performance test with a
remote cluster.
## Problem
In https://github.com/neondatabase/neon/pull/12513, the new code was
implemented to retry 404 errors caused by the replication lag. However,
this implemented the new logic, making the script more complicated,
while we have an existing one in `neon_api.py`.
## Summary of changes
The existing mechanism is used to retry 404 errors.
---------
Co-authored-by: Alexey Masterov <alexey.masterov@databricks.com>
## Problem
The communicator gRPC client currently uses bounded client/stream pools.
This can artificially constrain clients, especially after we remove
pipelining in #12584.
[Benchmarks](https://github.com/neondatabase/neon/pull/12583) show that
the cost of an idle server-side GetPage worker task is about 26 KB (2.5
GB for 100,000), so we can afford to scale out.
In the worst case, we'll degenerate to the current libpq state with one
stream per backend, but without the TCP connection overhead. In the
common case we expect significantly lower stream counts due to stream
sharing, driven e.g. by idle backends, LFC hits, read coalescing,
sharding (backends typically only talk to one shard at a time), etc.
Currently, Pageservers rarely serve more than 4000 backend connections,
so we have at least 2 orders of magnitude of headroom.
Touches #11735.
Requires #12584.
## Summary of changes
Remove the pool limits, and restructure the pools.
We still keep a separate bulk pool for Getpage batches of >4 pages (>32
KB), with fewer streams per connection. This reduces TCP-level
congestion and head-of-line blocking for non-bulk requests, and
concentrates larger window sizes on a smaller set of
streams/connections, presumably reducing memory usage. Apart from this,
bulk requests don't have any latency penalty compared to other requests.
## Problem
We don't log the timeline ID when rolling ephemeral layers during
housekeeping.
Resolves [LKB-179](https://databricks.atlassian.net/browse/LKB-179)
## Summary of changes
Add a span with timeline ID when calling `maybe_freeze_ephemeral_layer`
from the housekeeping loop.
We don't instrument the function itself, since future callers may not
have a span including the tenant_id already, but we don't want to
duplicate the tenant_id for these spans.
## Problem
The communicator gRPC client currently attempts to pipeline GetPage
requests from multiple callers onto the same gRPC stream. This has a
number of issues:
* Head-of-line blocking: the request may block on e.g. layer download or
LSN wait, delaying the next request.
* Cancellation: we can't easily cancel in-progress requests (e.g. due to
timeout or backend termination), so it may keep blocking the next
request (even its own retry).
* Complex stream scheduling: picking a stream becomes harder/slower, and
additional Tokio tasks and synchronization is needed for stream
management.
Touches #11735.
Requires #12579.
## Summary of changes
This patch removes pipelining of gRPC stream requests, and instead
prefers to scale out the number of streams to achieve the same
throughput. Stream scheduling has been rewritten, and mostly follows the
same pattern as the client pool with exclusive acquisition by a single
caller.
[Benchmarks](https://github.com/neondatabase/neon/pull/12583) show that
the cost of an idle server-side GetPage worker task is about 26 KB (2.5
GB for 100,000), so we can afford to scale out.
This has a number of advantages:
* It (mostly) eliminates head-of-line blocking (except at the TCP
level).
* Cancellation becomes trivial, by closing the stream.
* Stream scheduling becomes significantly simpler and cheaper.
* Individual callers can still use client-side batching for pipelining.
## Problem
The new communicator gRPC client has significantly worse Pagebench
performance than a basic gRPC client. We need to find out why.
## Summary of changes
Add a `pagebench --profile` flag which takes a client CPU profile of the
benchmark and writes a flamegraph to `profile.svg`.
## Problem
It can take 3x the idle timeout to reap a channel. We have to wait for
the idle timeout to trigger first for the stream, then the client, then
the channel.
Touches #11735.
## Summary of changes
Reap empty channels immediately, and rely indirectly on the
channel/stream timeouts.
This can still lead to 2x the idle timeout for streams (first stream
then client), but that's okay -- if the stream closes abruptly (e.g. due
to timeout or error) we want to keep the client around in the pool for a
while.
## Problem
gRPC client retries currently include pool acquisition under the
per-attempt timeout. If pool acquisition is slow (e.g. full pool), this
will cause spurious timeout warnings, and the caller will lose its place
in the pool queue.
Touches #11735.
## Summary of changes
Makes several improvements to retries and related logic:
* Don't include pool acquisition time under request timeouts.
* Move attempt timeouts out of `Retry` and into the closure.
* Make `Retry` configurable, move constants into main module.
* Don't backoff on the first retry, and reduce initial/max backoffs to
5ms and 5s respectively.
* Add `with_retries` and `with_timeout` helpers.
* Add slow logging for pool acquisition, and a `warn_slow` counterpart
to `log_slow`.
* Add debug logging for requests and responses at the client boundary.
## Problem
For the communicator scheduling policy, we need to understand the
server-side cost of idle gRPC streams.
Touches #11735.
## Summary of changes
Add an `idle-streams` benchmark to `pagebench` which opens a large
number of idle gRPC GetPage streams.
## Problem
When refreshing cancellation data we resend the entire value again just
to reset the TTL, which causes unnecessary load in proxy, on network and
possibly on redis side.
## Summary of changes
* Switch from using SET with full value to using EXPIRE to reset TTL.
* Add a tiny delay between retries to prevent busy loop.
* Shorten CancelKeyOp variants: drop redundant suffix.
* Retry SET when EXPIRE failed.
## Problem
When a connection terminates its maintain_cancel_key task keeps running
until the CANCEL_KEY_REFRESH sleep finishes and then it triggers another
cancel key TTL refresh before exiting.
## Summary of changes
* Check for cancellation while sleeping and interrupt sleep.
* If cancelled, break the loop, don't send a refresh cmd.
## Problem
We don't validate the validity of the `new_sk_set` before starting the
migration. It is validated later, so the migration to an invalid
safekeeper set will fail anyway. But at this point we might already
commited an invalid `new_sk_set` to the database and there is no `abort`
command yet (I ran into this issue in neon_local and ruined the timeline
:)
- Part of https://github.com/neondatabase/neon/issues/11669
## Summary of changes
- Add safekeeper count and safekeeper duplication checks before starting
the migration
- Test that we validate the `new_sk_set` before starting the migration
- Add `force` option to the `TimelineSafekeeperMigrateRequest` to
disable not-mandatory checks
Serialize query row responses directly into JSON. Some of this code
should be using the `json::value_as_object/list` macros, but I've
avoided it for now to minimize the size of the diff.
## Problem
To store cancellation data we send two commands to redis because the
redis server version doesn't support HSET with EX. Also, HSET is not
really needed.
## Summary of changes
* Replace the HSET + EXPIRE command pair with one SET .. EX command.
* Replace HGET with GET.
* Leave a workaround for old keys set with HSET.
* Replace some anyhow errors with specific errors to surface the
WRONGTYPE error from redis.
# TLDR
Problem-I is a bug fix. The rest are no-ops.
## Problem I
Page server checks image layer creation based on the elapsed time but
this check depends on the current logical size, which is only computed
on shard 0. Thus, for non-0 shards, the check will be ineffective and
image creation will never be done for idle tenants.
## Summary of changes I
This PR fixes the problem by simply removing the dependency on current
logical size.
## Summary of changes II
This PR adds a timeout when calling page server to split shard to make
sure SC does not wait for the API call forever. Currently the PR doesn't
adds any retry logic because it's not clear whether page server shard
split can be safely retried if the existing operation is still ongoing
or left the storage in a bad state. Thus it's better to abort the whole
operation and restart.
## Problem III
`test_remote_failures` requires PS to be compiled in the testing mode.
For PS in dev/staging, they are compiled without this mode.
## Summary of changes III
Remove the restriction and also increase the number of total failures
allowed.
## Summary of changes IV
remove test on PS getpage http route.
---------
Co-authored-by: Chen Luo <chen.luo@databricks.com>
Co-authored-by: Yecheng Yang <carlton.yang@databricks.com>
Co-authored-by: Vlad Lazar <vlad@neon.tech>
## Problem
close LKB-162
close https://github.com/neondatabase/cloud/issues/30665, related to
https://github.com/neondatabase/cloud/issues/29434
We see a lot of errors like:
```
2025-05-22T23:06:14.928959Z ERROR compaction_loop{tenant_id=? shard_id=0304}:run:gc_compact_timeline{timeline_id=?}: error applying 4 WAL records 35/DC0DF0B8..3B/E43188C0 (8119 bytes) to key 000000067F0000400500006027000000B9D0, from base image with LSN 0/0 to reconstruct page image at LSN 61/150B9B20 n_attempts=0: apply_wal_records
Caused by:
0: read walredo stdout
1: early eof
```
which is an acceptable form of error and we should downgrade it to
warning.
## Summary of changes
walredo error during gc-compaction is expected when the data below the
gc horizon does not contain a full key history. This is possible in some
rare cases of gc that is only able to remove data in the middle of the
history but not all earlier history when a full keyspace gets deleted.
Signed-off-by: Alex Chi Z <chi@neon.tech>
Fixes [LKB-61](https://databricks.atlassian.net/browse/LKB-61):
`test_timeline_archival_chaos` being flaky with storcon error `Requested
tenant is missing`.
When a tenant migration is ongoing, and the attach request has been sent
to the new location, but the attach hasn't finished yet, it is possible
for the pageserver to return a 412 precondition failed HTTP error on
timeline deletion, because it is being sent to the new location already.
That one we would previously log via sth like:
```
ERROR request{method=DELETE path=/v1/tenant/1f544a11c90d1afd7af9b26e48985a4e/timeline/32818fb3ebf07cb7f06805429d7dee38 request_id=c493c04b-7f33-46d2-8a65-aac8a5516055}: Error processing HTTP request: InternalServerError(Error deleting timeline 32
818fb3ebf07cb7f06805429d7dee38 on 1f544a11c90d1afd7af9b26e48985a4e on node 2 (localhost): pageserver API: Precondition failed: Requested tenant is missing
```
This patch changes that and makes us return a more reasonable resource
unavailable error. Not sure how scalable this is with tenants with a
large number of shards, but that's a different discussion (we'd probably
need a limited amount of per-storcon retries).
example
[link](https://neon-github-public-dev.s3.amazonaws.com/reports/pr-12398/15981821532/index.html#/testresult/e7785dfb1238d92f).
Update the WSS estimate before acquring the lock, so that we don't need
to hold the lock for so long. That seems safe to me, see added comment.
I was planning to do this with the new rust-based communicator
implementation anyway, but it might help a little with the current C
implementation too. And more importantly, having this as a separate PR
gives us a chance to review this aspect independently.
## Problem
Canceelation requires redis, redis required control-plane.
## Summary of changes
Make redis for cancellation not require control plane.
Add instructions for setting up redis locally.
Split the functions into two: one internal function to calculate the
estimate, and another (two functions) to expose it as SQL functions.
This is in preparation of adding new communicator implementation. With
that, the SQL functions will dispatch the call to the old or new
implementation depending on which is being used.
This is a no-op for the neon deployment
* Introduce the concept image consistent lsn: of the largest LSN below
which all pages have been redone successfully
* Use the image consistent LSN for forced image layer creations
* Optionally expose the image consistent LSN via the timeline describe
HTTP endpoint
* Add a sharded timeline describe endpoint to storcon
---------
Co-authored-by: Chen Luo <chen.luo@databricks.com>
## Problem
We have a `safekeeper_migrate` handler, but no subcommand in
`storcon_cli`. Same for `/:timeline_id/locate` for identifying current
set of safekeepers.
- Closes: https://github.com/neondatabase/neon/issues/12395
## Summary of changes
- Add `timeline-safekeeper-migrate` and `timeline-locate` subcommands to
`storcon_cli`
On December 8th, 2023, an engineering escalation (INC-110) was opened
after it was found that BYPASSRLS was being applied to all roles.
PR that introduced the issue:
https://github.com/neondatabase/neon/pull/5657
Subsequent commit on main:
ad99fa5f03
NOBYPASSRLS and INHERIT are the defaults for a Postgres role, but
because it isn't easy to know if a Postgres cluster is affected by the
issue, we need to keep the migration around for a long time, if not
indefinitely, so any cluster can be fixed.
Branching is the gift that keeps on giving...
Signed-off-by: Tristan Partin <tristan.partin@databricks.com>
Signed-off-by: Tristan Partin <tristan.partin@databricks.com>
## Problem
With gRPC `GetPageRequest` batches, we'll have non-trivial
fragmentation/reassembly logic in several places of the stack
(concurrent reads, shard splits, LFC hits, etc). If we included the
block numbers with the pages in `GetPageResponse` we could have better
verification and observability that the final responses are correct.
Touches #11735.
Requires #12480.
## Summary of changes
Add a `Page` struct with`block_number` for `GetPageResponse`, along with
the `RelTag` for completeness, and verify them in the rich gRPC client.
## Problem
Part of LKB-379
The pageserver connstrings are updated in the postmaster and then
there's a hook to propagate it to the shared memory of all backends.
However, the shard stripe doesn't. This would cause problems during
shard splits:
* the compute has active reads/writes
* shard split happens and the cplane applies the new config (pageserver
connstring + stripe size)
* pageserver connstring will be updated immediately once the postmaster
receives the SIGHUP, and it will be copied over the the shared memory of
all other backends.
* stripe size is a normal GUC and we don't have special handling around
that, so if any active backend has ongoing txns the value won't be
applied.
* now it's possible for backends to issue requests based on the wrong
stripe size; what's worse, if a request gets cached in the prefetch
buffer, it will get stuck forever.
## Summary of changes
To make sure it aligns with the current default in storcon.
Signed-off-by: Alex Chi Z <chi@neon.tech>
## Problem
`GetPageRequest::request_id` is supposed to be a unique ID for a
request. It's not, because we may retry the request using the same ID.
This causes assertion failures and confusion.
Touches #11735.
Requires #12480.
## Summary of changes
Extend the request ID with a retry attempt, and handle it in the gRPC
client and server.
## Problem
One PG tenant may write too fast and overwhelm the PS. The other tenants
sharing the same PSs will get very little bandwidth.
We had one experiment that two tenants sharing the same PSs. One tenant
runs a large ingestion that delivers hundreds of MB/s while the other
only get < 10 MB/s.
## Summary of changes
Rate limit how fast PG can generate WALs. The default is -1. We may
scale the default value with the CPU count. Need to run some experiments
to verify.
## How is this tested?
CI.
PGBench. No limit first. Then set to 1 MB/s and you can see the tps
drop. Then reverted the change and tps increased again.
pgbench -i -s 10 -p 55432 -h 127.0.0.1 -U cloud_admin -d postgres
pgbench postgres -c 10 -j 10 -T 6000000 -P 1 -b tpcb-like -h 127.0.0.1
-U cloud_admin -p 55432
progress: 33.0 s, 986.0 tps, lat 10.142 ms stddev 3.856 progress: 34.0
s, 973.0 tps, lat 10.299 ms stddev 3.857 progress: 35.0 s, 1004.0 tps,
lat 9.939 ms stddev 3.604 progress: 36.0 s, 984.0 tps, lat 10.183 ms
stddev 3.713 progress: 37.0 s, 998.0 tps, lat 10.004 ms stddev 3.668
progress: 38.0 s, 648.9 tps, lat 12.947 ms stddev 24.970 progress: 39.0
s, 0.0 tps, lat 0.000 ms stddev 0.000 progress: 40.0 s, 0.0 tps, lat
0.000 ms stddev 0.000 progress: 41.0 s, 0.0 tps, lat 0.000 ms stddev
0.000 progress: 42.0 s, 0.0 tps, lat 0.000 ms stddev 0.000 progress:
43.0 s, 0.0 tps, lat 0.000 ms stddev 0.000 progress: 44.0 s, 0.0 tps,
lat 0.000 ms stddev 0.000 progress: 45.0 s, 0.0 tps, lat 0.000 ms stddev
0.000 progress: 46.0 s, 0.0 tps, lat 0.000 ms stddev 0.000 progress:
47.0 s, 0.0 tps, lat 0.000 ms stddev 0.000 progress: 48.0 s, 0.0 tps,
lat 0.000 ms stddev 0.000 progress: 49.0 s, 347.3 tps, lat 321.560 ms
stddev 1805.633 progress: 50.0 s, 346.8 tps, lat 9.898 ms stddev 3.809
progress: 51.0 s, 0.0 tps, lat 0.000 ms stddev 0.000 progress: 52.0 s,
0.0 tps, lat 0.000 ms stddev 0.000 progress: 53.0 s, 0.0 tps, lat 0.000
ms stddev 0.000 progress: 54.0 s, 0.0 tps, lat 0.000 ms stddev 0.000
progress: 55.0 s, 0.0 tps, lat 0.000 ms stddev 0.000 progress: 56.0 s,
0.0 tps, lat 0.000 ms stddev 0.000 progress: 57.0 s, 0.0 tps, lat 0.000
ms stddev 0.000 progress: 58.0 s, 0.0 tps, lat 0.000 ms stddev 0.000
progress: 59.0 s, 0.0 tps, lat 0.000 ms stddev 0.000 progress: 60.0 s,
0.0 tps, lat 0.000 ms stddev 0.000 progress: 61.0 s, 0.0 tps, lat 0.000
ms stddev 0.000 progress: 62.0 s, 0.0 tps, lat 0.000 ms stddev 0.000
progress: 63.0 s, 494.5 tps, lat 276.504 ms stddev 1853.689 progress:
64.0 s, 488.0 tps, lat 20.530 ms stddev 71.981 progress: 65.0 s, 407.8
tps, lat 9.502 ms stddev 3.329 progress: 66.0 s, 0.0 tps, lat 0.000 ms
stddev 0.000 progress: 67.0 s, 0.0 tps, lat 0.000 ms stddev 0.000
progress: 68.0 s, 504.5 tps, lat 71.627 ms stddev 397.733 progress: 69.0
s, 371.0 tps, lat 24.898 ms stddev 29.007 progress: 70.0 s, 541.0 tps,
lat 19.684 ms stddev 24.094 progress: 71.0 s, 342.0 tps, lat 29.542 ms
stddev 54.935
Co-authored-by: Haoyu Huang <haoyu.huang@databricks.com>
After https://github.com/neondatabase/neon/pull/12240 we observed
issues in our go code as `ComputeStatus` is not stateless, thus doesn't
deserialize as string.
```
could not check compute activity: json: cannot unmarshal object into Go struct field
ComputeState.status of type computeclient.ComputeStatus
```
- Fix this by splitting this status into two.
- Update compute OpenApi spec to reflect changes to `/terminate` in
previous PR
## Problem
If we have catalog update AND a pageserver migration batched in a single
spec, we will not be able to apply the spec (running the SQL) because
the compute is not attached to the right pageserver and we are not able
to read anything if we don't pick up the latest pageserver connstring.
This is not a case for now because cplane always schedules shard split /
pageserver migrations with `skip_pg_catalog_updates` (I suppose).
Context:
https://databricks.slack.com/archives/C09254R641L/p1752163559259399?thread_ts=1752160163.141149&cid=C09254R641L
With this fix, backpressure will likely not be able to affect
reconfigurations.
## Summary of changes
Do `pg_reload_conf` before we apply specs in SQL.
---------
Signed-off-by: Alex Chi Z <chi@neon.tech>
## Problem
We need to benchmark the rich gRPC client
`client_grpc::PageserverClient` against the basic, no-frills
`page_api::Client` to determine how much overhead it adds.
Touches #11735.
Requires #12476.
## Summary of changes
Add a `pagebench --rich-client` parameter to use
`client_grpc::PageserverClient`. Also adds a compression parameter to
the client.
## Problem
Sometimes we run out of free ports in `PortDistributor`. This affects
particularly failed tests that we rerun automatically up to 3 times
(which makes it use up to 3x more ports)
## Summary of changes
- Cycle over the range of ports to reuse freed ports from previous tests
Ref: LKB-62
## Problem
The communicator gRPC client must support changing the shard map on
splits.
Touches #11735.
Requires #12476.
## Summary of changes
* Wrap the shard set in a `ArcSwap` to allow swapping it out.
* Add a new `ShardSpec` parameter struct to pass validated shard info to
the client.
* Add `update_shards()` to change the shard set. In-flight requests are
allowed to complete using the old shards.
* Restructure `get_page` to use a stable view of the shard map, and
retry errors at the top (pre-split) level to pick up shard map changes.
* Also marks `tonic::Status::Internal` as non-retryable, so that we can
use it for client-side invariant checks without continually retrying
these.
# TLDR
All changes are no-op except some metrics.
## Summary of changes I
### Pageserver
Added a new global counter metric
`pageserver_pagestream_handler_results_total` that categorizes
pagestream request results according to their outcomes:
1. Success
2. Internal errors
3. Other errors
Internal errors include:
1. Page reconstruction error: This probably indicates a pageserver
bug/corruption
2. LSN timeout error: Could indicate overload or bugs with PS's ability
to reach other components
3. Misrouted request error: Indicates bugs in the Storage Controller/HCC
Other errors include transient errors that are expected during normal
operation or errors indicating bugs with other parts of the system
(e.g., malformed requests, errors due to cancelled operations during PS
shutdown, etc.)
## Summary of changes II
This PR adds a pageserver endpoint and its counterpart in storage
controller to list visible size of all tenant shards. This will be a
prerequisite of the tenant rebalance command.
## Problem III
We need a way to download WAL
segments/layerfiles from S3 and replay WAL records. We cannot access
production S3 from our laptops directly, and we also can't transfer any
user data out of production systems for GDPR compliance, so we need
solutions.
## Summary of changes III
This PR adds a couple of tools to support the debugging
workflow in production:
1. A new `pagectl download-remote-object` command that can be used to
download remote storage objects assuming the correct access is set up.
## Summary of changes IV
This PR adds a command to list all visible delta and image layers from
index_part. This is useful to debug compaction issues as index_part
often contain a lot of covered layers due to PITR.
---------
Co-authored-by: William Huang <william.huang@databricks.com>
Co-authored-by: Chen Luo <chen.luo@databricks.com>
Co-authored-by: Vlad Lazar <vlad@neon.tech>
## Problem
close LKB-253
## Summary of changes
404 for timeline requests could happen when the tenant is intended to be
on a pageserver but not attached yet. This patch adds handling for the
lease request. In the future, we should extend this handling to more
operations.
---------
Signed-off-by: Alex Chi Z <chi@neon.tech>
## Problem
Safekeeper and pageserver metrics collection might time out. We've seen
this in both hadron and neon.
## Summary of changes
This PR moves metrics collection in PS/SK to the background so that we
will always get some metrics, despite there may be some delays. Will
leave it to the future work to reduce metrics collection time.
---------
Co-authored-by: Chen Luo <chen.luo@databricks.com>
## Problem
The gRPC client pools don't reap idle resources.
Touches #11735.
Requires #12475.
## Summary of changes
Reap idle pool resources (channels/clients/streams) after 3 minutes of
inactivity.
Also restructure the `StreamPool` to use a mutex rather than atomics for
synchronization, for simplicity. This will be optimized later.
This PR introduces a `image_creation_timeout` to page servers so that we
can force the image creation after a certain period. This is set to 1
day on dev/staging for now, and will rollout to production 1/2 weeks
later.
Majority of the PR are boilerplate code to add the new knob. Specific
changes of the PR are:
1. During L0 compaction, check if we should force a compaction if
min(LSN) of all delta layers < force_image_creation LSN.
2. During image creation, check if we should force a compaction if the
image's LSN < force_image_creation LSN and there are newer deltas with
overlapping key ranges.
3. Also tweaked the check image creation interval to make sure we honor
image_creation_timeout.
Vlad's note: This should be a no-op. I added an extra PS config for the
large timeline
threshold to enable this.
---------
Co-authored-by: Chen Luo <chen.luo@databricks.com>
When a function is owned by a superuser (bootstrap user or otherwise),
we consider it safe to run it. Only a superuser could have installed it,
typically from CREATE EXTENSION script: we trust the code to execute.
## Problem
This is intended to solve running pg_graphql Event Triggers
graphql_watch_ddl and graphql_watch_drop which are executing the secdef
function graphql.increment_schema_version().
## Summary of changes
Allow executing Event Trigger function owned by a superuser and with
SECURITY DEFINER properties. The Event Trigger code runs with superuser
privileges, and we consider that it's fine.
---------
Co-authored-by: Tristan Partin <tristan.partin@databricks.com>
There are a couple of places that call `CompactionError::is_cancel` but
don't check the `::Other` variant via downcasting for root cause being
cancellation.
The only place that does it is `log_compaction_error`.
It's sad we have to do it, but, until we get around cleaning up all the
culprits,
a step forward is to unify the behavior so that all places that inspect
a
`CompactionError` for cancellation reason follow the same behavior.
Thus, this PR ...
- moves the downcasting checks against the `::Other` variant from
`log_compaction_error` into `is_cancel()` and
- enforces via type system that `.is_cancel()` is used to check whether
a CompactionError is due to cancellation (matching on the
`CompactionError::ShuttingDown` will cause a compile-time error).
I don't think there's a _serious_ case right now where matching instead
of using `is_cancel` causes problems.
The worst case I could find is the circuit breaker and
`compaction_failed`,
which don't really matter if we're shutting down the timeline anyway.
But it's unaesthetic and might cause log/alert noise down the line,
so, this PR fixes that at least.
Refs
- https://databricks.atlassian.net/browse/LKB-182
- slack conversation about this PR:
https://databricks.slack.com/archives/C09254R641L/p1751284317955159
## Problem
close LKB-199
## Summary of changes
We always return the error as 500 to the cplane if a LSN lease request
fails. This cause issues for the cplane as they don't retry on 500. This
patch correctly passes through the error and assign the error code so
that cplane can know if it is a retryable error. (TODO: look at the
cplane code and learn the retry logic).
Note that this patch does not resolve LKB-253 -- we need to handle not
found error separately in the lsn lease path, like wait until the tenant
gets attached, or return 503 so that cplane can retry.
---------
Signed-off-by: Alex Chi Z <chi@neon.tech>
Change the unreliable storage wrapper to fail by probability when there
are more failure attempts left.
Co-authored-by: Yecheng Yang <carlton.yang@databricks.com>
## Problem
Test `test_branch_creation_before_gc` is flaky in the internal repo.
Pageserver sometimes lags behind write LSN. When we call GC it might not
reach the LSN we try to create the branch at yet.
## Summary of changes
- Wait till flush lsn on pageserver reaches the latest LSN before
calling GC.
## Problem
GetPage bulk requests such as prefetches and vacuum can head-of-line
block foreground requests, causing increased latency.
Touches #11735.
Requires #12469.
## Summary of changes
* Use dedicated channel/client/stream pools for bulk GetPage requests.
* Use lower concurrency but higher queue depth for bulk pools.
* Make pool limits configurable.
* Require unbounded client pool for stream pool, to avoid accidental
starvation.
## Problem
Follow up of #12400
## Summary of changes
We didn't set remote_size_mb to Some when initialized so it never gets
computed :(
Also added a new API to force refresh the properties.
Signed-off-by: Alex Chi Z <chi@neon.tech>
## Problem
Due to a lag in replication, we sometimes cannot get the parent branch
definition just after completion of the Public API restore call. This
leads to the test failures.
https://databricks.atlassian.net/browse/LKB-279
## Summary of changes
The workaround is implemented. Now test retries up to 30 seconds,
waiting for the branch definition to appear.
---------
Co-authored-by: Alexey Masterov <alexey.masterov@databricks.com>
## Problem
As discovered in https://github.com/neondatabase/neon/issues/12394,
test_multiple_subscription_branching generates skewed data distribution,
that leads to test failures when the unevenly filled last table receives
even more data.
for table t0: pub_res = (42001,), sub_res = (42001,)
for table t1: pub_res = (29001,), sub_res = (29001,)
for table t2: pub_res = (21001,), sub_res = (21001,)
for table t3: pub_res = (21001,), sub_res = (21001,)
for table t4: pub_res = (1711001,), sub_res = (1711001,)
## Summary of changes
Fix the name of the workload parameter to generate data as expected.
## Problem
The rich gRPC Pageserver client needs to split GetPage batches that
straddle multiple shards.
Touches #11735.
Requires #12462.
## Summary of changes
Adds a `GetPageSplitter` which splits `GetPageRequest` that span
multiple shards, and then reassembles the responses. Dispatches
per-shard requests in parallel.
## Problem
See [Slack
Channel](https://databricks.enterprise.slack.com/archives/C091LHU6NNB)
Dropping connection without resetting prefetch state can cause
request/response mismatch.
And lack of check response correctness in communicator_prefetch_lookupv
can cause data corruption.
## Summary of changes
1. Validate response before assignment to prefetch slot.
2. Consume prefetch requests before sending any other requests.
---------
Co-authored-by: Kosntantin Knizhnik <konstantin.knizhnik@databricks.com>
Co-authored-by: Konstantin Knizhnik <knizhnik@neon.tech>
The `--timelines-onto-safekeepers` flag is very consequential in the
sense that it controls every single timeline creation. However, we don't
have any automatic insight whether enabling the option will break things
or not.
The main way things can break is by misconfigured safekeepers, say they
are marked as paused in the storcon db. The best input so far we can
obtain via manually connecting via storcon_cli and listing safekeepers,
but this is cumbersome and manual so prone to human error.
So at storcon startup, do a simulated "test creation" in which we call
`timelines_onto_safekeepers` with the configuration provided to us, and
print whether it was successful or not. No actual timeline is created,
and nothing is written into the storcon db. The heartbeat info will not
have reached us at that point yet, but that's okay, because we still
fall back to safekeepers that don't have any heartbeat.
Also print some general scheduling policy stats on initial safekeeper
load.
Part of #11670.
## Problem
For the communicator, we need a rich Pageserver gRPC client.
Touches #11735.
Requires #12434.
## Summary of changes
This patch adds an initial rich Pageserver gRPC client. It supports:
* Sharded tenants across multiple Pageservers.
* Pooling of connections, clients, and streams for efficient resource
use.
* Concurrent use by many callers.
* Internal handling of GetPage bidirectional streams, with pipelining
and error handling.
* Automatic retries.
* Observability.
The client is still under development. In particular, it needs GetPage
batch splitting, shard map updates, and performance optimization. This
will be addressed in follow-up PRs.
The only differentiated handling of it is for `is_critical`, which in
turn is a `matches!()` on several variants of the `enum
CollectKeyspaceError`
which is the value contained insided
`CompactionError::CollectKeyspaceError`.
This PR introduces a new error for `repartition()`, allowing its
immediate
callers to inspect it like `is_critical` did.
A drive-by fix is more precise classification of WaitLsnError::BadState
when mapping to `tonic::Status`.
refs
- https://databricks.atlassian.net/browse/LKB-182
## Problem
close LKB-209
## Summary of changes
- We should not allow lease creation below the applied gc cutoff.
- Also removed the condition for `AttachedSingle`. We should always
check the lease against the gc cutoff in all attach modes.
---------
Signed-off-by: Alex Chi Z <chi@neon.tech>
## Problem
We only trim the senders if we tried to send a message to them and
discovered that the channel is closed. This is problematic if the
pageserver keeps connecting while there's nothing to send back for the
shard. In this scenario we never trim down the senders list and can
panic due to the u8 limit.
## Summary of Changes
Trim down the dead senders before adding a new one.
Closes LKB-178
## Problem
We lost capability to explicitly disable the global eviction task (for
testing).
## Summary of changes
Add an `enabled` flag to `DiskUsageEvictionTaskConfig` to indicate
whether we should run the eviction job or not.
## Problem
The communicator will need gRPC channel/client/stream pools for
efficient reuse across many backends.
Touches #11735.
Requires #12396.
## Summary of changes
Adds three nested resource pools:
* `ChannelPool` for gRPC channels (i.e. TCP connections).
* `ClientPool` for gRPC clients (i.e. `page_api::Client`). Acquires
channels from `ChannelPool`.
* `StreamPool` for gRPC GetPage streams. Acquires clients from
`ClientPool`.
These are minimal functional implementations that will need further
improvements and performance optimization. However, the overall
structure is expected to be roughly final, so reviews should focus on
that.
The pools are not yet in use, but will form the foundation of a rich
gRPC Pageserver client used by the communicator (see #12462). This PR
also adds the initial crate scaffolding for that client.
See doc comments for details.
# TLDR
All changes are no-op except
1. publishing additional metrics.
2. problem VI
## Problem I
It has come to my attention that the Neon Storage Controller doesn't
correctly update its "observed" state of tenants previously associated
with PSs that has come back up after a local data loss. It would still
think that the old tenants are still attached to page servers and won't
ask more questions. The pageserver has enough information from the
reattach request/response to tell that something is wrong, but it
doesn't do anything about it either. We need to detect this situation in
production while I work on a fix.
(I think there is just some misunderstanding about how Neon manages
their pageserver deployments which got me confused about all the
invariants.)
## Summary of changes I
Added a `pageserver_local_data_loss_suspected` gauge metric that will be
set to 1 if we detect a problematic situation from the reattch response.
The problematic situation is when the PS doesn't have any local tenants
but received a reattach response containing tenants.
We can set up an alert using this metric. The alert should be raised
whenever this metric reports non-zero number.
Also added a HTTP PUT
`http://pageserver/hadron-internal/reset_alert_gauges` API on the
pageserver that can be used to reset the gauge and the alert once we
manually rectify the situation (by restarting the HCC).
## Problem II
Azure upload is 3x slower than AWS. -> 3x slower ingestion.
The reason for the slower upload is that Azure upload in page server is
much slower => higher flush latency => higher disk consistent LSN =>
higher back pressure.
## Summary of changes II
Use Azure put_block API to uploads a 1 GB layer file in 8 blocks in
parallel.
I set the put_block block size to be 128 MB by default in azure config.
To minimize neon changes, upload function passes the layer file path to
the azure upload code through the storage metadata. This allows the
azure put block to use FileChunkStreamRead to stream read from one
partition in the file instead of loading all file data in memory and
split it into 8 128 MB chunks.
## How is this tested? II
1. rust test_real_azure tests the put_block change.
3. I deployed the change in azure dev and saw flush latency reduces from
~30 seconds to 10 seconds.
4. I also did a bunch of stress test using sqlsmith and 100 GB TPCDS
runs.
## Problem III
Currently Neon limits the compaction tasks as 3/4 * CPU cores. This
limits the overall compaction throughput and it can easily cause
head-of-the-line blocking problems when a few large tenants are
compacting.
## Summary of changes III
This PR increases the limit of compaction tasks as `BG_TASKS_PER_THREAD`
(default 4) * CPU cores. Note that `CONCURRENT_BACKGROUND_TASKS` also
limits some other tasks `logical_size_calculation` and `layer eviction`
. But compaction should be the most frequent and time-consuming task.
## Summary of changes IV
This PR adds the following PageServer metrics:
1. `pageserver_disk_usage_based_eviction_evicted_bytes_total`: captures
the total amount of bytes evicted. It's more straightforward to see the
bytes directly instead of layers.
2. `pageserver_active_storage_operations_count`: captures the active
storage operation, e.g., flush, L0 compaction, image creation etc. It's
useful to visualize these active operations to get a better idea of what
PageServers are spending cycles on in the background.
## Summary of changes V
When investigating data corruptions, it's useful to search the base
image and all WAL records of a page up to an LSN, i.e., a breakdown of
GetPage@LSN request. This PR implements this functionality with two
tools:
1. Extended `pagectl` with a new command to search the layer files for a
given key up to a given LSN from the `index_part.json` file. The output
can be used to download the files from S3 and then search the file
contents using the second tool.
Example usage:
```
cargo run --bin pagectl index-part search --tenant-id 09b99ea3239bbb3b2d883a59f087659d --timeline-id 7bedf4a6995baff7c0421ff9aebbcdab --path ~/Downloads/corruption/index_part.json-0000000c-formatted --key 000000067F000080140000802100000D61BD --lsn 70C/BF3D61D8
```
Example output:
```
tenants/09b99ea3239bbb3b2d883a59f087659d-0304/timelines/7bedf4a6995baff7c0421ff9aebbcdab/000000067F0000801400000B180000000002-000000067F0000801400008028000002FEFF__000007089F0B5381-0000070C7679EEB9-0000000c
tenants/09b99ea3239bbb3b2d883a59f087659d-0304/timelines/7bedf4a6995baff7c0421ff9aebbcdab/000000000000000000000000000000000000-000000067F0000801400008028000002F3F1__000006DD95B6F609-000006E2BA14C369-0000000c
tenants/09b99ea3239bbb3b2d883a59f087659d-0304/timelines/7bedf4a6995baff7c0421ff9aebbcdab/000000067F0000801400000B180000000002-000000067F000080140000802100001B0973__000006D33429F539-000006DD95B6F609-0000000c
tenants/09b99ea3239bbb3b2d883a59f087659d-0304/timelines/7bedf4a6995baff7c0421ff9aebbcdab/000000067F0000801400000B180000000002-000000067F00008014000080210000164D81__000006C6343B2D31-000006D33429F539-0000000b
tenants/09b99ea3239bbb3b2d883a59f087659d-0304/timelines/7bedf4a6995baff7c0421ff9aebbcdab/000000067F0000801400000B180000000002-000000067F0000801400008021000017687B__000006BA344FA7F1-000006C6343B2D31-0000000b
tenants/09b99ea3239bbb3b2d883a59f087659d-0304/timelines/7bedf4a6995baff7c0421ff9aebbcdab/000000067F0000801400000B180000000002-000000067F00008014000080210000165BAB__000006AD34613D19-000006BA344FA7F1-0000000b
tenants/09b99ea3239bbb3b2d883a59f087659d-0304/timelines/7bedf4a6995baff7c0421ff9aebbcdab/000000067F0000801400000B180000000002-000000067F00008014000080210000137A39__0000069F34773461-000006AD34613D19-0000000b
tenants/09b99ea3239bbb3b2d883a59f087659d-0304/timelines/7bedf4a6995baff7c0421ff9aebbcdab/000000067F000080140000802100000D4000-000000067F000080140000802100000F0000__0000069F34773460-0000000b
```
2. Added a unit test to search the layer file contents. It's not
implemented part of `pagectl` because it depends on some test harness
code, which can only be used by unit tests.
Example usage:
```
cargo test --package pageserver --lib -- tenant::debug::test_search_key --exact --nocapture -- --tenant-id 09b99ea3239bbb3b2d883a59f087659d --timeline-id 7bedf4a6995baff7c0421ff9aebbcdab --data-dir /Users/chen.luo/Downloads/corruption --key 000000067F000080140000802100000D61BD --lsn 70C/BF3D61D8
```
Example output:
```
# omitted image for brievity
delta: 69F/769D8180: will_init: false, "OgAAALGkuwXwYp12nwYAAECGAAASIqLHAAAAAH8GAAAUgAAAIYAAAL1hDQD/DLGkuwUDAAAAEAAWAA=="
delta: 69F/769CB6D8: will_init: false, "PQAAALGkuwXotZx2nwYAABAJAAAFk7tpACAGAH8GAAAUgAAAIYAAAL1hDQD/CQUAEAASALExuwUBAAAAAA=="
```
## Problem VI
Currently when page service resolves shards from page numbers, it
doesn't fully support the case that the shard could be split in the
middle. This will lead to query failures during the tenant split for
either commit or abort cases (it's mostly for abort).
## Summary of changes VI
This PR adds retry logic in `Cache::get()` to deal with shard resolution
errors more gracefully. Specifically, it'll clear the cache and retry,
instead of failing the query immediately. It also reduces the internal
timeout to make retries faster.
The PR also fixes a very obvious bug in
`TenantManager::resolve_attached_shard` where the code tries to cache
the computed the shard number, but forgot to recompute when the shard
count is different.
---------
Co-authored-by: William Huang <william.huang@databricks.com>
Co-authored-by: Haoyu Huang <haoyu.huang@databricks.com>
Co-authored-by: Chen Luo <chen.luo@databricks.com>
Co-authored-by: Vlad Lazar <vlad.lazar@databricks.com>
Co-authored-by: Vlad Lazar <vlad@neon.tech>
This patch tightens up `page_api::Client`. It's mostly superficial
changes, but also adds a new constructor that takes an existing gRPC
channel, for use with the communicator connection pool.
## Problem
Some feature flags are used heavily on the critical path and we want the
"get feature flag" operation as cheap as possible.
## Summary of changes
Add a `test_remote_size_flag` as an example of such flags. In the
future, we can use macro to generate all those fields. The flag is
updated in the housekeeping loop. The retrieval of the flag is simply
reading an atomic flag.
---------
Signed-off-by: Alex Chi Z <chi@neon.tech>
## Problem
The goal of this code was to test out if resetting the broker
subscription helps alleviate the issues we've been seeing in staging.
Looks like it did the trick. However, the original version was too
eager.
## Summary of Changes
Only reset the stream when:
* we are waiting for WAL
* there's no connection candidates lined up
* we're not already connected to a safekeeper
The only call stack that can emit the `::AlreadyRunning` variant is
```
-> iteration_inner
-> iteration
-> compaction_iteration
-> compaction_loop
-> start_background_loops
```
And on that call stack, the only differentiated handling of it is its
invocations of
`log_compaction_error -> CompactionError::is_cancel()`, which returns
`true` for
`::AlreadyRunning`.
I think the condition of `AlreadyRunning` is severe; it really shouldn't
happen.
So, this PR starts treating it as something that is to be logged at
`ERROR` / `WARN`
level, depending on the `degrate_to_warning` argument to
`log_compaction_error`.
refs
- https://databricks.atlassian.net/browse/LKB-182
## Problem
Grafana Alloy in cluster mode seems to send duplicate "seconds" scrape
URL parameters
when one of its instances is disrupted.
## Summary of changes
Temporarily accept duplicate parameters as long as their value is
identical.
Looks can be deceiving: the match blocks in
`maybe_trip_compaction_breaker`
and at the end of `compact_with_options` seem like differentiated error
handling, but in reality, these branches are unreachable at runtime
because the only source of `CompactionError::Offload` within the
compaction code is at the end of `Tenant::compaction_iteration`.
We can simply map offload cancellation to CompactionError::Cancelled and
all other offload errors to ::Other, since there's no differentiated
handling for them in the compaction code.
Also, the OffloadError::RemoteStorage variant has no differentiated
handling, but was wrapping the remote storage anyhow::Error in a
`anyhow(thiserror(anyhow))` sandwich. This PR removes that variant,
mapping all RemoteStorage errors to `OffloadError::Other`.
Thereby, the sandwich is gone and we will get a proper anyhow backtrace
to the remote storage error location if when we debug-print the
OffloadError (or the CompactionError if we map it to that).
refs
- https://databricks.atlassian.net/browse/LKB-182
- the observation that there's no need for differentiated handling of
CompactionError::Offload was made in
https://databricks.slack.com/archives/C09254R641L/p1751286453930269?thread_ts=1751284317.955159&cid=C09254R641L
Before this PR, macOS builds would get clippy warning
```
warning: `tokio_epoll_uring::thread_local_system` does not refer to an existing function
```
The reason is that the `thread_local_system` function is only defined on
Linux.
Add `allow-invalid = true` to make macOS clippy pass, and manually test
that on Linux builds, clippy still fails when we use it.
refs
- https://databricks.slack.com/archives/C09254R641L/p1751917655527099
Co-authored-by: Christian Schwarz <Christian Schwarz>
## Problem
Deletion process does not calculate preferred nodes correctly - it
doesn't consider current tenant-shard layout among all pageservers.
## Summary of changes
- Added a schedule context calculation for node deletion
Co-authored-by: Aleksandr Sarantsev <aleksandr.sarantsev@databricks.com>
The introduction of the default lease deadline feature 9 months ago made
it so
that after PS restart, `.timeline_gc()` calls in Python tests are no-ops
for 10 minute after pageserver startup: the `gc_iteration()` bails with
`Skipping GC because lsn lease deadline is not reached`.
I did some impact analysis in the following PR. About 30 Python tests
are affected:
- https://github.com/neondatabase/neon/pull/12411
Rust tests that don't explicitly enable periodic GC or invoke GC
manually
are unaffected because we disable periodic GC by default in
the `TenantHarness`'s tenant config.
Two tests explicitly did `start_paused=true` + `tokio::time::advance()`,
but it would add cognitive and code bloat to each existing and future
test case that uses TenantHarness if we took that route.
So, this PR sets the default lease deadline feature in both Python
and Rust tests to zero by default. Tests that test the feature were
thus identified by failing the test:
- Python test `test_readonly_node_gc` + `test_lsn_lease_size`
- Rust test `test_lsn_lease`.
To accomplish the above, I changed the code that computes the initial
lease
deadline to respect the pageserver.toml's default tenant config, which
it didn't before (and I would consider a bug). The Python test harness
and the Rust TenantHarness test harness then simply set the default
tenant
config field to zero.
Drive-by:
- `test_lsn_lease_size` was writing a lot of data unnecessarily; reduce
the amount and speed up the test
refs
- PR that introduced default lease deadline:
https://github.com/neondatabase/neon/pull/9055/files
- fixes https://databricks.atlassian.net/browse/LKB-92
---------
Co-authored-by: Christian Schwarz <Christian Schwarz>
## Problem
Extension tests were previously run sequentially, resulting in
unnecessary wait time and underutilization of available CPU cores.
## Summary of changes
Tests are now executed in a customizable number of parallel threads
using separate database branches. This reduces overall test time by
approximately 50% (e.g., on my laptop, parallel test lasts 173s, while
sequential one lasts 340s) and increases the load on the pageserver,
providing better test coverage.
---------
Co-authored-by: Alexander Bayandin <alexander@neon.tech>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Co-authored-by: Alexey Masterov <alexey.masterov@databricks.com>
## Problem
Pageserver now writes errors in the log during the safekeeper migration.
Some errors are added to allowed errors, but "timeline not found in
global map" is not.
- Will be properly fixed in
https://github.com/neondatabase/neon/issues/12191
## Summary of changes
Add "timeline not found in global map" error in a list of allowed errors
in `test_safekeeper_migration_simple`
## Problem
Test is flaky due to the following warning in the logs:
```
Keeping extra secondaries: can't determine which of [NodeId(1), NodeId(2)] to remove (some nodes offline?)
```
Some nodes being offline is expected behavior in this test.
## Summary of changes
- Added `Keeping extra secondaries` to the list of allowed errors
- Improved logging for better debugging experience
Co-authored-by: Aleksandr Sarantsev <aleksandr.sarantsev@databricks.com>
## Problem
The endpoint filter cache is still unused because it's not yet reliable
enough to be used. It only consumes a lot of memory.
## Summary of changes
Remove the code. Needs a new design.
neondatabase/cloud#30634
See #11992 and #11961 for some examples of usecases.
This introduces a JSON serialization lib, designed for more flexibility
than serde_json offers.
## Dynamic construction
Sometimes you have dynamic values you want to serialize, that are not
already in a serde-aware model like a struct or a Vec etc. To achieve
this with serde, you need to implement a lot of different traits on a
lot of different new-types. Because of this, it's often easier to
give-in and pull all the data into a serde-aware model
(serde_json::Value or some intermediate struct), but that is often not
very efficient.
This crate allows full control over the JSON encoding without needing to
implement any extra traits. Just call the relevant functions, and it
will guarantee a correctly encoded JSON value.
## Async construction
Similar to the above, sometimes the values arrive asynchronously. Often
collecting those values in memory is more expensive than writing them as
JSON, since the overheads of `Vec` and `String` is much higher, however
there are exceptions.
Serializing to JSON all in one go is also more CPU intensive and can
cause lag spikes, whereas serializing values incrementally spreads out
the CPU load and reduces lag.
## Problem
We don't log gRPC request errors on the server.
Touches #11728.
## Summary of changes
Automatically log non-OK gRPC response statuses in the observability
middleware, and add corresponding logging for the `get_pages` stream.
Also adds the peer address and gRPC method to the gRPC tracing span.
Example output:
```
2025-07-02T20:18:16.813718Z WARN grpc:pageservice{peer=127.0.0.1:56698 method=CheckRelExists tenant_id=c7b45faa1924b1958f05c5fdee8b0d04 timeline_id=4a36ee64fd2f97781b9dcc2c3cddd51b shard_id=0000}: request failed with NotFound: Tenant c7b45faa1924b1958f05c5fdee8b0d04 not found
```
## Problem
Names are not consistent between safekeeper migration RFC and the actual
implementation.
It's not used anywhere in production yet, so it's safe to rename. We
don't need to worry about backward compatibility.
- Follow up on https://github.com/neondatabase/neon/pull/12432
## Summary of changes
- rename term -> last_log_term in TimelineMembershipSwitchResponse
- add missing fields to TimelineMembershipSwitchResponse in python
## Problem
The safekeeper migration code/logic slightly diverges from the initial
RFC. This PR aims to address these differences.
- Part of https://github.com/neondatabase/neon/issues/12192
## Summary of changes
- Adjust the RFC to reflect that we implemented the safekeeper
reconciler with in-memory queue.
- Add `sk_set_notified_generation` field to the `timelines` table in the
RFC to address the "finish migration atomically" problem.
- Describe how we are going to make the timeline migration handler fully
retriable with in-memory reconciler queue.
- Unify type/field/method names in the code and RFC.
- Fix typos
This doesn't do anything interesting yet, but demonstrates linking Rust
code to the neon Postgres extension, so that we can review and test
drive just the build process changes independently.
Mostly a revert of https://github.com/neondatabase/neon/pull/11851 and
https://github.com/neondatabase/neon/pull/12330 .
Christian suggested reverting his PR to fix the issue
https://github.com/neondatabase/neon/issues/12369 .
Alternatives considered:
1. I have originally wanted to introduce cancellation tokens to
`RequestContext`, but in the end I gave up on them because I didn't find
a select-free way of preventing
`test_layer_download_cancelled_by_config_location` from hanging.
Namely if I put a select around the `get_or_maybe_download` invocation
in `get_values_reconstruct_data`, it wouldn't hang, but if I put it
around the `download_init_and_wait` invocation in
`get_or_maybe_download`, the test would still hang. Not sure why, even
though I made the attached child function of the `RequestContext` create
a child token.
2. Introduction of a `download_cancel` cancellation token as a child of
a timeline token, putting it into `RemoteTimelineClient` together with
the main token, and then putting it into the whole
`RemoteTimelineClient` read path.
3. Greater refactorings, like to make cancellation tokens follow a DAG
structure so you can have tokens cancelled either by say timeline
shutting down or a request ending. It doesn't just represent an effort
that we don't have the engineering budget for, it also causes
interesting questions like what to do about batching (do you cancel the
entire request if only some requests get cancelled?).
We might see a reemergence of
https://github.com/neondatabase/neon/issues/11762, but given that we
have https://github.com/neondatabase/neon/pull/11853 and
https://github.com/neondatabase/neon/pull/12376 now, it is possible that
it will not come back. Looking at some code, it might actually fix the
locations where the error pops up. Let's see.
---------
Co-authored-by: Christian Schwarz <christian@neon.tech>
- Add ComputeSpec flag `offload_lfc_interval_seconds` controlling
whether LFC should be offloaded to endpoint storage. Default value
(None) means "don't offload".
- Add glue code around it for `neon_local` and integration tests.
- Add `autoprewarm` mode for `test_lfc_prewarm` testing
`offload_lfc_interval_seconds` and `autoprewarm` flags in conjunction.
- Rename `compute_ctl_lfc_prewarm_requests_total` and
`compute_ctl_lfc_offload_requests_total` to
`compute_ctl_lfc_prewarms_total`
and `compute_ctl_lfc_offloads_total` to reflect we count prewarms and
offloads, not `compute_ctl` requests of those.
Don't count request in metrics if there is a prewarm/offload already
ongoing.
https://github.com/neondatabase/cloud/issues/19011
Resolves: https://github.com/neondatabase/cloud/issues/30770
Don't build walproposer-lib as a separate job. It only takes a few
seconds, after you have built all its dependencies.
Don't cache the Neon Pg extensions in the per-postgres-version caches.
This is in preparation for the communicator project, which will
introduce Rust parts to the Neon Pg extension, which complicates the
build process. With that, the 'make neon-pg-ext' step requires some of
the Rust bits to be built already, or it will build them on the spot,
which in turn requires all the Rust sources to be present, and we don't
want to repeat that part for each Postgres version anyway. To prepare
for that, rely on "make all" to build the neon extension and the rust
bits in the correct order instead. Building the neon extension doesn't
currently take very long anyway after you have built Postgres itself, so
you don't gain much by caching it. See
https://github.com/neondatabase/neon/pull/12266.
Add an explicit "rustup update" step to update the toolchain. It's not
strictly necessary right now, because currently "make all" will only
invoke "cargo build" once and the race condition described in the
comment doesn't happen. But prepare for the future.
To further simplify the build, get rid of the separate 'build-postgres'
jobs too, and just build Postgres as a step in the main job. That makes
the overall workflow run longer, because we no longer build all the
postgres versions in parallel (although you still get intra-runner
parallelism thanks to `make -j`), but that's acceptable. In the
cache-hit case, it might even be a little faster because there is less
overhead from launching jobs, and in the cache-miss case, it's maybe
5-10 minutes slower altogether.
---------
Co-authored-by: Alexander Bayandin <alexander@neon.tech>
## Problem
The current deletion operation is synchronous and blocking, which is
unsuitable for potentially long-running tasks like. In such cases, the
standard HTTP request-response pattern is not a good fit.
## Summary of Changes
- Added new `storcon_cli` commands: `NodeStartDelete` and
`NodeCancelDelete` to initiate and cancel deletion asynchronously.
- Added corresponding `storcon` HTTP handlers to support the new
start/cancel deletion flow.
- Introduced a new type of background operation: `Delete`, to track and
manage the deletion process outside the request lifecycle.
---------
Co-authored-by: Aleksandr Sarantsev <aleksandr.sarantsev@databricks.com>
## Problem
I suspect that the pageservers get stuck on receiving broker updates.
## Summary of changes
This is a an opportunistic (staging only) patch that resets the
susbscription
stream if it's been idle for a while. This won't go to prod in this
form.
I'll revert or update it before Friday.
## Problem
DEBUG_LOCAL_COMPARE mode allows to detect data corruption.
But it requires rebuild of neon extension (and so requires special
image) and significantly slowdown execution because always fetch pages
from page server.
## Summary of changes
Introduce new GUC `neon.debug_compare_local`, accepting the following
values: " none", "prefetch", "lfc", "all" (by default it is definitely
disabled).
In mode less than "all", neon SMGR will not fetch page from PS if it is
found in local caches.
Co-authored-by: Konstantin Knizhnik <knizhnik@neon.tech>
## Problem
The background installed extensions worker relied on `interval.tick()`
to go to sleep for a period of time. This can lead to bugs due to the
interval being updated at the end of the loop as the first tick is
[instantaneous](https://docs.rs/tokio/latest/tokio/time/struct.Interval.html#method.tick).
## Summary of changes
Changed it to a `tokio::time::sleep` to prevent this issue. Now it puts
the thread to sleep and only wakes up after the specified duration
## Problem
Better investigate memory usage during backpressure
## Summary of changes
Print open layer size if backpressure is activated
Signed-off-by: Alex Chi Z <chi@neon.tech>
## Problem
Ancestor detach from a previously detached parent when there were no
writes panics since it tries to upload the tombstone layer twice.
## Summary of Changes
If we're gonna copy the tombstone from the ancestor, don't bother
creating it.
Fixes https://github.com/neondatabase/neon/issues/12458
When deploying new safekeepers, we don't immediately want to send
traffic to them. Maybe they are not ready yet by the time the deploy
script is registering them with the storage controller.
For pageservers, the storcon solves the problem by not scheduling stuff
to them unless there has been a positive heartbeat response. We can't do
the same for safekeepers though, otherwise a single down safekeeper
would mean we can't create new timelines in smaller regions where there
is only three safekeepers in total.
So far we have created safekeepers as `pause` but this adds a manual
step to safekeeper deployment which is prone to oversight. We want
things to be automatted. So we introduce a new state `activating` that
acts just like `pause`, except that we automatically transition the
policy to `active` once we get a positive heartbeat from the safekeeper.
For `pause`, we always keep the safekeeper paused.
## Problem
Currently, if `storcon` (storage controller) reconciliations repeatedly
fail, the system will indefinitely freeze optimizations. This can result
in optimization starvation for several days until the reconciliation
issues are manually resolved. To mitigate this, we should detect
persistently failing reconciliations and exclude them from influencing
the optimization decision.
## Summary of Changes
- A tenant shard reconciliation is now considered "keep-failing" if it
fails 5 consecutive times. These failures are excluded from the
optimization readiness check.
- Added a new metric: `storage_controller_keep_failing_reconciles` to
monitor such cases.
- Added a warning log message when a reconciliation is marked as
"keep-failing".
---------
Co-authored-by: Aleksandr Sarantsev <aleksandr.sarantsev@databricks.com>
## TLDR
This PR is a no-op. The changes are disabled by default.
## Problem
I. Currently we don't have a way to detect disk I/O failures from WAL
operations.
II.
We observe that the offloader fails to upload a segment due to race
conditions on XLOG SWITCH and PG start streaming WALs. wal_backup task
continously failing to upload a full segment while the segment remains
partial on the disk.
The consequence is that commit_lsn for all SKs move forward but
backup_lsn stays the same. Then, all SKs run out of disk space.
III.
We have discovered SK bugs where the WAL offload owner cannot keep up
with WAL backup/upload to S3, which results in an unbounded accumulation
of WAL segment files on the Safekeeper's disk until the disk becomes
full. This is a somewhat dangerous operation that is hard to recover
from because the Safekeeper cannot write its control files when it is
out of disk space. There are actually 2 problems here:
1. A single problematic timeline can take over the entire disk for the
SK
2. Once out of disk, it's difficult to recover SK
IV.
Neon reports certain storage errors as "critical" errors using a marco,
which will increment a counter/metric that can be used to raise alerts.
However, this metric isn't sliced by tenant and/or timeline today. We
need the tenant/timeline dimension to better respond to incidents and
for blast radius analysis.
## Summary of changes
I.
The PR adds a `safekeeper_wal_disk_io_errors ` which is incremented when
SK fails to create or flush WALs.
II.
To mitigate this issue, we will re-elect a new offloader if the current
offloader is lagging behind too much.
Each SK makes the decision locally but they are aware of each other's
commit and backup lsns.
The new algorithm is
- determine_offloader will pick a SK. say SK-1.
- Each SK checks
-- if commit_lsn - back_lsn > threshold,
-- -- remove SK-1 from the candidate and call determine_offloader again.
SK-1 will step down and all SKs will elect the same leader again.
After the backup is caught up, the leader will become SK-1 again.
This also helps when SK-1 is slow to backup.
I'll set the reelect backup lag to 4 GB later. Setting to 128 MB in dev
to trigger the code more frequently.
III.
This change addresses problem no. 1 by having the Safekeeper perform a
timeline disk utilization check check when processing WAL proposal
messages from Postgres/compute. The Safekeeper now rejects the WAL
proposal message, effectively stops writing more WAL for the timeline to
disk, if the existing WAL files for the timeline on the SK disk exceeds
a certain size (the default threshold is 100GB). The disk utilization is
calculated based on a `last_removed_segno` variable tracked by the
background task removing WAL files, which produces an accurate and
conservative estimate (>= than actual disk usage) of the actual disk
usage.
IV.
* Add a new metric `hadron_critical_storage_event_count` that has the
`tenant_shard_id` and `timeline_id` as dimensions.
* Modified the `crtitical!` marco to include tenant_id and timeline_id
as additional arguments and adapted existing call sites to populate the
tenant shard and timeline ID fields. The `critical!` marco invocation
now increments the `hadron_critical_storage_event_count` with the extra
dimensions. (In SK there isn't the notion of a tenant-shard, so just the
tenant ID is recorded in lieu of tenant shard ID.)
I considered adding a separate marco to avoid merge conflicts, but I
think in this case (detecting critical errors) conflicts are probably
more desirable so that we can be aware whenever Neon adds another
`critical!` invocation in their code.
---------
Co-authored-by: Chen Luo <chen.luo@databricks.com>
Co-authored-by: Haoyu Huang <haoyu.huang@databricks.com>
Co-authored-by: William Huang <william.huang@databricks.com>
## Problem
integrating subzero requires a bit of refactoring. To make the
integration PR a bit more manageable, the refactoring is done in this
separate PR.
## Summary of changes
* move common types/functions used in sql_over_http to errors.rs and
http_util.rs
* add the "Local" auth backend to proxy (similar to local_proxy), useful
in local testing
* change the Connect and Send type for the http client to allow for
custom body when making post requests to local_proxy from the proxy
---------
Co-authored-by: Ruslan Talpa <ruslan.talpa@databricks.com>
## Problem
`keep_connection` does not exit, so it was never setting
`credentials_refreshed`.
## Summary of changes
Set `credentials_refreshed` to true when we first establish a
connection, and after we re-authenticate the connection.
## Problem
#12450
## Summary of changes
Instead of `#[arg(default_value_t = typed_default_value)]`, we use
`#[arg(default_value = "str that deserializes into the value")]`,
because apparently you can't convince clap to _not_ deserialize from the
Display implementation of an imported enum.
Health checks for pg-sni-router open a TCP connection and immediately
close it again. This is noisy. We will filter out any EOF errors on the
first message.
"acquired permit" debug log is incorrect since it logs when we timedout
as well. This fixes the debug log.
## Problem
When local proxy is configured with TLS, the certificate does not match
the endpoint string. This currently returns an error.
## Summary of changes
I don't think this code is necessary anymore, taking the prefix from the
hostname is good enough (and is equivalent to what `endpoint_sni` was
doing) and we ignore checking the domain suffix.
## Problem
Neon currently implements several features that guarantee high uptime of
compute nodes:
1. Storage high-availability (HA), i.e. each tenant shard has a
secondary pageserver location, so we can quickly switch over compute to
it in case of primary pageserver failure.
2. Fast compute provisioning, i.e. we have a fleet of pre-created empty
computes, that are ready to serve workload, so restarting unresponsive
compute is very fast.
3. Preemptive NeonVM compute provisioning in case of k8s node
unavailability.
This helps us to be well-within the uptime SLO of 99.95% most of the
time. Problems begin when we go up to multi-TB workloads and 32-64 CU
computes. During restart, compute looses all caches: LFC, shared
buffers, file system cache. Depending on the workload, it can take a lot
of time to warm up the caches, so that performance could be degraded and
might be even unacceptable for certain workloads. The latter means that
although current approach works well for small to
medium workloads, we still have to do some additional work to avoid
performance degradation after restart of large instances.
[Rendered
version](https://github.com/neondatabase/neon/blob/alexk/pg-prewarm-rfc/docs/rfcs/2025-03-17-compute-prewarm.md)
Part of https://github.com/neondatabase/cloud/issues/19011
## Problem
The billing team wants to change the billing events pipeline and use a
common events format in S3 buckets across different event producers.
## Summary of changes
Change the events storage format for billing events from JSON to NDJSON.
Also partition files by hours, rather than days.
Resolves: https://github.com/neondatabase/cloud/issues/29995
We would like to move towards an arena system for JSON encoding the
responses. This change pushes an "out" parameter into the pg-test to
json routines to make swapping in an arena system easier in the future.
(see #11992)
This additionally removes the redundant `column: &[Type]` argument, as
well as rewriting the pg_array parser.
---
I rewrote the pg_array parser since while making these changes I found
it hard to reason about. I went back to the specification and rewrote it
from scratch. There's 4 separate routines:
1. pg_array_parse - checks for any prelude (multidimensional array
ranges)
2. pg_array_parse_inner - only deals with the arrays themselves
3. pg_array_parse_item - parses a single item from the array, this might
be quoted, unquoted, or another nested array.
4. pg_array_parse_quoted - parses a quoted string, following the
relevant string escaping rules.
Don't print errors like:
```
Compaction failed 1 times, retrying in 2s: Failed to offload timeline: Unexpected offload error: Timeline deletion is already in progress
```
Print it at info log level instead.
https://github.com/neondatabase/cloud/issues/30666
## Problem
https://github.com/neondatabase/neon/pull/11712 changed how computes are
started in the test: the lsn is specified, making them read-only static
replicas. Lsn is `last_record_lsn` from pageserver. It works fine with
read-only branches (because their `last_record_lsn` is equal to
`start_lsn` and always valid). But with writable timelines, the
`last_record_lsn` on the pageserver might be stale.
Particularly in this test, after the `detach_branch` operation, the
tenant is reset on the pagesever. It leads to `last_record_lsn` going
back to `disk_consistent_lsn`, so basically rolling back some recent
writes.
If we start a primary compute, it will start at safekeepers' commit Lsn,
which is the correct one , and will wait till pageserver catches up with
this Lsn after reset.
- Closes: https://github.com/neondatabase/neon/issues/12365
## Summary of changes
- Start `primary` compute for writable timelines.
## Problem
Part of #11813
## Summary of changes
* Compute tenant remote size in the housekeeping loop.
* Add a new `TenantFeatureResolver` struct to cache the tenant-specific
properties.
* Evaluate feature flag based on the remote size.
---------
Signed-off-by: Alex Chi Z <chi@neon.tech>
## Problem
close https://github.com/neondatabase/neon/issues/11528
## Summary of changes
Gives us better observability of compaction progress.
- Image creation: num of partition processed / total partition
- Gc-compaction: index of the in the queue / total items for a full
compaction
- Shard ancestor compaction: layers to rewrite / total layers
Signed-off-by: Alex Chi Z <chi@neon.tech>
## Problem
L0 compaction currently holds the read lock for a long region while it
doesn't need to.
## Summary of changes
This patch reduces the one long contention region into 2 short ones:
gather the layers to compact at the beginning, and several short read
locks when querying the image coverage.
Co-Authored-By: Chen Luo
---------
Signed-off-by: Alex Chi Z <chi@neon.tech>
## Problem
This PR works on adding dynamic masking support for `anon` v2. It
currently only supports static masking.
## Summary of changes
Added a security definer function that sets the dynamic masking guc to
`true` with superuser permissions.
Added a security definer function that adds `anon` to
`session_preload_libraries` if it's not already present.
Related to: https://github.com/neondatabase/cloud/issues/20456
## Problem
Currently we do not have metrics for autovacuum.
## Summary of changes
Added a metric that extracts the top 5 DBs with oldest mxid and frozen
xid. Tables that were vacuumed recently should have younger value (or
younger age).
Related Issue: https://github.com/neondatabase/cloud/issues/27296
## Problem
#8843
User initiated sql queries are being classified as "postgres" errors,
whereas they're really user errors.
## Summary of changes
Classify user-initiated postgres errors as user errors if they are
related to a sql query that we ran on their behalf. Do not log those
errors.
Add support to transport syslogs over TLS. Since TLS params essentially
require passing host and port separately, add a boolean flag to the
configuration template and also use the same `action` format for
plaintext logs. This allows seamless transition.
The plaintext host:port is picked from `AUDIT_LOGGING_ENDPOINT` (as
earlier) and from `AUDIT_LOGGING_TLS_ENDPOINT`. The TLS host:port is
used when defined and non-empty.
`remote_endpoint` is split separately to hostname and port as required
by `omfwd` module.
Also the address parsing and config content generation are split to more
testable functions with basic tests added.
## Problem
We persist safekeeper host/port in the storcon DB after
https://github.com/neondatabase/neon/pull/11712, so the storcon fails to
ping safekeepers in the compatibility tests, where we start the cluster
from the snapshot.
PR also adds some small code improvements related to the test failure.
- Closes: https://github.com/neondatabase/neon/issues/12339
## Summary of changes
- Update safekeeper ports in the storcon DB when starting the neon from
the dir (snapshot)
- Fail the response on all not-success codes (e.g. 3xx). Should not
happen, but just to be more safe.
- Add `neon_previous/` to .gitignore to make it easier to run compat
tests.
- Add missing EXPORT to the instruction for running compat tests
## Problem
Previously, the background worker that collects the list of installed
extensions across DBs had a timeout set to 1 hour. This cause a problem
with computes that had a `suspend_timeout` > 1 hour as this collection
was treated as activity, preventing compute shutdown.
Issue: https://github.com/neondatabase/cloud/issues/30147
## Summary of changes
Passing the `suspend_timeout` as part of the `ComputeSpec` so that any
updates to this are taken into account by the background worker and
updates its collection interval.
If a hardlink operation inside `detach_ancestor` fails due to the layer
already existing, we delete the layer to make sure the source is one we
know about, and then retry.
But we deleted the wrong file, namely, the one we wanted to use as the
source of the hardlink. As a result, the follow up hard link operation
failed. Our PR corrects this mistake.
## Problem
We don't notify cplane about safekeeper membership change yet. Without
the notification the compute needs to know all the safekeepers on the
cluster to be able to speak to them. Change notifications will allow to
avoid it.
- Closes: https://github.com/neondatabase/neon/issues/12188
## Summary of changes
- Implement `notify_safekeepers` method in `ComputeHook`
- Notify cplane about safekeepers in `safekeeper_migrate` handler.
- Update the test to make sure notifications work.
## Out of scope
- There is `cplane_notified_generation` field in `timelines` table in
strocon's database. It's not needed now, so it's not updated in the PR.
Probably we can remove it.
- e2e tests to make sure it works with a production cplane
## Problem
The gRPC API does not provide LSN leases.
## Summary of changes
* Add LSN lease support to the gRPC API.
* Use gRPC LSN leases for static computes with `grpc://` connstrings.
* Move `PageserverProtocol` into the `compute_api::spec` module and
reuse it.
## Problem
Location config changes can currently result in changes to the shard
identity. Such changes will cause data corruption, as seen with #12217.
Resolves#12227.
Requires #12377.
## Summary of changes
Assert that the shard identity does not change on location config
updates and on (re)attach.
This is currently asserted with `critical!`, in case it misfires in
production. Later, we should reject such requests with an error and turn
this into a proper assertion.
Implicit mapping to an `anyhow::Error` when we do `?` is discouraged
because tooling to find those places isn't great.
As a drive-by, also make SplitImageLayerWriter::new infallible and sync.
I think we should also make ImageLayerWriter::new completely lazy,
then `BatchLayerWriter:new` infallible and async.
Reverts neondatabase/neon#11663 and
https://github.com/neondatabase/neon/pull/11265/
Step Security is not yet approved by Databricks team, in order to
prevent issues during Github org migration, I'll revert this PR to use
the previous action instead of Step Security maintained action.
## Problem
Similarly to #12217, the following endpoints may result in a stripe size
mismatch between the storage controller and Pageserver if an unsharded
tenant has a different stripe size set than the default. This can lead
to data corruption if the tenant is later manually split without
specifying an explicit stripe size, since the storage controller and
Pageserver will apply different defaults. This commonly happens with
tenants that were created before the default stripe size was changed
from 32k to 2k.
* `PUT /v1/tenant/config`
* `PATCH /v1/tenant/config`
These endpoints are no longer in regular production use (they were used
when cplane still managed Pageserver directly), but can still be called
manually or by tests.
## Summary of changes
Retain the current shard parameters when updating the location config in
`PUT | PATCH /v1/tenant/config`.
Also opportunistically derive `Copy` for `ShardParameters`.
## Problem
Some of the design decisions in PR #12256 were influenced by the
requirements of consistency tests. These decisions introduced
intermediate logic that is no longer needed and should be cleaned up.
## Summary of Changes
- Remove the `feature("testing")` flag related to
`kick_secondary_download`.
- Set the default value of `kick_secondary_download` back to false,
reflecting the intended production behavior.
Co-authored-by: Aleksandr Sarantsev <aleksandr.sarantsev@databricks.com>
## Problem
- Closes: https://github.com/neondatabase/neon/issues/12298
## Summary of changes
- Set `timeline_safekeeper_count` in `neon_local` if we have less than 3
safekeepers
- Remove `cfg!(feature = "testing")` code from
`safekeepers_for_new_timeline`
- Change `timeline_safekeeper_count` type to `usize`
## Problem
`compute_ctl` should support gRPC base backups.
Requires #12111.
Requires #12243.
Touches #11926.
## Summary of changes
Support `grpc://` connstrings for `compute_ctl` base backups.
## Problem
The problem has been well described in already-commited PR #11853.
tl;dr: BufferedWriter is sensitive to cancellation, which the previous
approach was not.
The write path was most affected (ingest & compaction), which was mostly
fixed in #11853:
it introduced `PutError` and mapped instances of `PutError` that were
due to cancellation of underlying buffered writer into
`CreateImageLayersError::Cancelled`.
However, there is a long tail of remaining errors that weren't caught by
#11853 that result in `CompactionError::Other`s, which we log with great
noise.
## Solution
The stack trace logging for CompactionError::Other added in #11853
allows us to chop away at that long tail using the following pattern:
- look at the stack trace
- from leaf up, identify the place where we incorrectly map from the
distinguished variant X indicating cancellation to an `anyhow::Error`
- follow that anyhow further up, ensuring it stays the same anyhow all
the way up in the `CompactionError::Other`
- since it stayed one anyhow chain all the way up, root_cause() will
yield us X
- so, in `log_compaction_error`, add an additional `downcast_ref` check
for X
This PR specifically adds checks for
- the flush task cancelling (FlushTaskError, BlobWriterError)
- opening of the layer writer (GateError)
That should cover all the reports in issues
- https://github.com/neondatabase/cloud/issues/29434
- https://github.com/neondatabase/neon/issues/12162
## Refs
- follow-up to #11853
- fixup of / fixes https://github.com/neondatabase/neon/issues/11762
- fixes https://github.com/neondatabase/neon/issues/12162
- refs https://github.com/neondatabase/cloud/issues/29434
We don't have cancellation support for timeline deletions. In other
words, timeline deletion might still go on in an older generation while
we are attaching it in a newer generation already, because the
cancellation simply hasn't reached the deletion code.
This has caused us to hit a situation with offloaded timelines in which
the timeline was in an unrecoverable state: always returning an accepted
response, but never a 404 like it should be.
The detailed description can be found in
[here](https://github.com/neondatabase/cloud/issues/30406#issuecomment-3008667859)
(private repo link).
TLDR:
1. we ask to delete timeline on old pageserver/generation, starts
process in background
2. the storcon migrates the tenant to a different pageserver.
- during attach, the pageserver still finds an index part, so it adds it
to `offloaded_timelines`
4. the timeline deletion finishes, removing the index part in S3
5. there is a retry of the timeline deletion endpoint, sent to the new
pageserver location. it is bound to fail however:
- as the index part is gone, we print `Timeline already deleted in
remote storage`.
- the problem is that we then return an accepted response code, and not
a 404.
- this confuses the code calling us. it thinks the timeline is not
deleted, so keeps retrying.
- this state never gets recovered from until a reset/detach, because of
the `offloaded_timelines` entry staying there.
This is where this PR fixes things: if no index part can be found, we
can safely assume that the timeline is gone in S3 (it's the last thing
to be deleted), so we can remove it from `offloaded_timelines` and
trigger a reupload of the manifest. Subsequent retries will pick that
up.
Why not improve the cancellation support? It is a more disruptive code
change, that might have its own risks. So we don't do it for now.
Fixes https://github.com/neondatabase/cloud/issues/30406
Mainly for general readability. Some notable changes:
- Postgres can be built without the rest of the repository, and in
particular without any of the Rust bits. Some CI scripts took advantage
of that, so let's make that more explicit by separating those parts.
Also add an explicit comment about that in the new postgres.mk file.
- Add a new PG_INSTALL_CACHED variable. If it's set, `make all` and
other top-Makefile targets skip checking if Postgres is up-to-date. This
is also to be used in CI scripts that build and cache Postgres as
separate steps. (It is currently only used in the macos walproposer-lib
rule, but stay tuned for more.)
- Introduce a POSTGRES_VERSIONS variable that lists all supported
PostgreSQL versions. Refactor a few Makefile rules to use that.
## Problem
In large oltp test run
https://github.com/neondatabase/neon/actions/runs/15905488707/job/44859116742
we see that the `Benchmark database maintenance` step is skipped in all
3 strategy variants, however it should be executed in two.
This is due to treating the `test_maintenance` boolean type in the
strategy in the condition of the `Benchmark database maintenance` step
## Summary of changes
Use a boolean condition instead of a string comparison
## Test run from this pull request branch
https://github.com/neondatabase/neon/actions/runs/15923605412
## Problem
When resolving a shard during a split we might have multiple attached
shards with the old shard count (i.e. not all of them are marked in
progress and ignored). Hence, we can compute the desired shard number
based on the old shard count and misroute the request.
## Summary of Changes
Recompute the desired shard every time the shard count changes during
the iteration
# Problem
In #12335 I moved the `authenticate` method outside of the
`connect_to_compute` loop. This triggered [e2e tests to become
flaky](https://github.com/neondatabase/cloud/pull/30533). This
highlighted an edge case we forgot to consider with that change.
When we connect to compute, the compute IP might be cached. This cache
hit might however be stale. Because we can't validate the IP is
associated with a specific compute-id☨, we will succeed the
connect_to_compute operation and fail when it comes to password
authentication☨☨. Before the change, we were invalidating the cache and
triggering wake_compute if the authentication failed.
Additionally, I noticed some faulty logic I introduced 1 year ago
https://github.com/neondatabase/neon/pull/8141/files#diff-5491e3afe62d8c5c77178149c665603b29d88d3ec2e47fc1b3bb119a0a970afaL145-R147
☨ We can when we roll out TLS, as the certificate common name includes
the compute-id.
☨☨ Technically password authentication could pass for the wrong compute,
but I think this would only happen in the very very rare event that the
IP got reused **and** the compute's endpoint happened to be a
branch/replica.
# Solution
1. Fix the broken logic
2. Simplify cache invalidation (I don't know why it was so convoluted)
3. Add a loop around connect_to_compute + authenticate to re-introduce
the wake_compute invalidation we accidentally removed.
I went with this approach to try and avoid interfering with
https://github.com/neondatabase/neon/compare/main...cloneable/proxy-pglb-connect-compute-split.
The changes made in commit 3 will move into `handle_client_request` I
suspect,
## Problem
Some pageservers hit `max_size_entries` limit in staging with only ~25
MiB storage used by basebackup cache. The limit is too strict. It should
be safe to relax it.
- Part of https://github.com/neondatabase/cloud/issues/29353
## Summary of changes
- Increase the default `max_size_entries` from 1000 to 10000
## Problem
Druing shard splits we shut down the remote client early and allow the
parent shard to keep ingesting data. While ingesting data, the wal
receiver task may wait for the current flush to complete in order to
apply backpressure. Notifications are delivered via
`Timeline::layer_flush_done_tx`.
When the remote client was being shut down the flush loop exited
whithout delivering a notification. This left
`Timeline::wait_flush_completion` hanging indefinitely which blocked the
shutdown of the wal receiver task, and, hence, the shard split.
## Summary of Changes
Deliver a final notification when the flush loop is shutting down
without the timeline cancel cancellation token having fired. I tried
writing a test for this, but got stuck in failpoint hell and decided
it's not worth it.
`test_sharding_autosplit`, which reproduces this reliably in CI, passed
with the proposed fix in
https://github.com/neondatabase/neon/pull/12304.
Closes https://github.com/neondatabase/neon/issues/12060
Make the safekeeper `pull_timeline` endpoint support timelines that
haven't had any writes yet. In the storcon managed sk timelines world,
if a safekeeper goes down temporarily, the storcon will schedule a
`pull_timeline` call. There is no guarantee however that by when the
safekeeper is online again, there have been writes to the timeline yet.
The `snapshot` endpoint gives an error if the timeline hasn't had
writes, so we avoid calling it if `timeline_start_lsn` indicates a
freshly created timeline.
Fixes#11422
Part of #11670
## Problem
gRPC base backups do not use the base backup cache.
Touches https://github.com/neondatabase/neon/issues/11728.
## Summary of changes
Integrate gRPC base backups with the base backup cache.
Also fixes a bug where the base backup cache did not differentiate
between primary/replica base backups (at least I think that's a bug?).
## Problem
In our infra config, we have to split server_api_key and other fields in
two files: the former one in the sops file, and the latter one in the
normal config. It creates the situation that we might misconfigure some
regions that it only has part of the fields available, causing
storcon/pageserver refuse to start.
## Summary of changes
Allow PostHog config to have part of the fields available. Parse it
later.
Signed-off-by: Alex Chi Z <chi@neon.tech>
## Problem
Basebackup cache now uses unbounded channel for prepare requests. In
theory it can grow large if the cache is hung and does not process the
requests.
- Part of https://github.com/neondatabase/cloud/issues/29353
## Summary of changes
- Replace an unbounded channel with a bounded one, the size is
configurable.
- Add `pageserver_basebackup_cache_prepare_queue_size` to observe the
size of the queue.
- Refactor a bit to move all metrics logic to `basebackup_cache.rs`
I was looking at how we could expose our proxy config as toml again, and
as I was writing out the schema format, I noticed some cruft in our CLI
args that no longer seem to be in use.
The redis change is the most complex, but I am pretty sure it's sound.
Since https://github.com/neondatabase/cloud/pull/15613 cplane longer
publishes to the global redis instance.
## Problem
If LFC generation is changed then `lfc_readv_select` will return -1 but
pages are still marked as available in bitmap.
## Summary of changes
Update bitmap after generation check.
Co-authored-by: Kosntantin Knizhnik <konstantin.knizhnik@databricks.com>
## Problem
Fix for https://github.com/neondatabase/neon/pull/12324
## Summary of changes
Need `serde(default)` to allow this field not present in the config,
otherwise there will be a config deserialization error.
---------
Signed-off-by: Alex Chi Z <chi@neon.tech>
## Problem
gRPC base backups use gRPC compression. However, this has two problems:
* Base backup caching will cache compressed base backups (making gRPC
compression pointless).
* Tonic does not support varying the compression level, and zstd default
level is 10% slower than gzip fastest level.
Touches https://github.com/neondatabase/neon/issues/11728.
Touches https://github.com/neondatabase/cloud/issues/29353.
## Summary of changes
This patch adds a gRPC parameter `BaseBackupRequest::compression`
specifying the compression algorithm. It also moves compression into
`send_basebackup_tarball` to reduce code duplication.
A follow-up PR will integrate the base backup cache with gRPC.
## Summary
A design for a storage system that allows storage of files required to
make
Neon's Endpoints have a better experience at or after a reboot.
## Motivation
Several systems inside PostgreSQL (and Neon) need some persistent
storage for
optimal workings across reboots and restarts, but still work without.
Examples are the cumulative statistics file in `pg_stat/global.stat`,
`pg_stat_statements`' `pg_stat/pg_stat_statements.stat`, and
`pg_prewarm`'s
`autoprewarm.blocks`. We need a storage system that can store and manage
these files for each Endpoint.
[GH rendered
file](https://github.com/neondatabase/neon/blob/MMeent/rfc-unlogged-file/docs/rfcs/040-Endpoint-Persistent-Unlogged-Files-Storage.md)
Part of https://github.com/neondatabase/cloud/issues/24225
## Problem
part of https://github.com/neondatabase/neon/issues/11813
## Summary of changes
It costs $$$ to directly retrieve the feature flags from the pageserver.
Therefore, this patch adds new APIs to retrieve the spec from the
storcon and updates it via pageserver.
* Storcon retrieves the feature flag and send it to the pageservers.
* If the feature flag gets updated outside of the normal refresh loop of
the pageserver, pageserver won't fetch the flags on its own as long as
the last updated time <= refresh_period.
Signed-off-by: Alex Chi Z <chi@neon.tech>
## Problem
https://github.com/neondatabase/cloud/issues/30539
If the current leader cancels the `call` function, then it has removed
the jobs from the queue, but will never finish sending the responses.
Because of this, it is not cancellation safe.
## Summary of changes
Document these functions as not cancellation safe. Move cancellation of
the queued jobs into the queue itself.
## Alternatives considered
1. We could spawn the task that runs the batch, since that won't get
cancelled.
* This requires `fn call(self: Arc<Self>)` or `fn call(&'static self)`.
2. We could add another scopeguard and return the requests back to the
queue.
* This requires that requests are always retry safe, and also requires
requests to be `Clone`.
## Problem
While working more on TLS to compute, I realised that Console Redirect
-> pg-sni-router -> compute would break if channel binding was set to
prefer. This is because the channel binding data would differ between
Console Redirect -> pg-sni-router vs pg-sni-router -> compute.
I also noticed that I actually disabled channel binding in #12145, since
`connect_raw` would think that the connection didn't support TLS.
## Summary of changes
Make sure we specify the channel binding.
Make sure that `connect_raw` can see if we have TLS support.
In `test_layer_download_cancelled_by_config_location`, we simulate hung
downloads via the `before-downloading-layer-stream-pausable` failpoint.
Then, we cancel a timeline via the `location_config` endpoint.
With the new default as of
https://github.com/neondatabase/neon/pull/11712, we would be creating
the timeline on safekeepers regardless if there have been writes or not,
and it turns out the test relied on the timeline not existing on
safekeepers, due to a cancellation bug:
* as established before, the test makes the read path hang
* the timeline cancellation function first cancels the walreceiver, and
only then cancels the timeline's token
* `WalIngest::new` is requesting a checkpoint, which hits the read path
* at cancellation time, we'd be hanging inside the read, not seeing the
cancellation of the walreceiver
* the test would time out due to the hang
This is probably also reproducible in the wild when there is S3
unavailabilies or bottlenecks. So we thought that it's worthwhile to fix
the hang issue. The approach chosen in the end involves the
`tokio::select` macro.
In PR 11712, we originally punted on the test due to the hang and opted
it out from the new default, but now we can use the new default.
Part of https://github.com/neondatabase/neon/issues/12299
pgaudit can spam logs due to all the monitoring that we do. Logs from
these connections are not necessary for HIPPA compliance, so we can stop
logging from those connections.
Part-of: https://github.com/neondatabase/cloud/issues/29574
Signed-off-by: Tristan Partin <tristan@neon.tech>
This makes it possible for the compiler to validate that a match block
matched all PostgreSQL versions we support.
## Problem
We did not have a complete picture about which places we had to test
against PG versions, and what format these versions were: The full PG
version ID format (Major/minor/bugfix `MMmmbb`) as transfered in
protocol messages, or only the Major release version (`MM`). This meant
type confusion was rampant.
With this change, it becomes easier to develop new version-dependent
features, by making type and niche confusion impossible.
## Summary of changes
Every use of `pg_version` is now typed as either `PgVersionId` (u32,
valued in decimal `MMmmbb`) or PgMajorVersion (an enum, with a value for
every major version we support, serialized and stored like a u32 with
the value of that major version)
---------
Co-authored-by: Arpad Müller <arpad-m@users.noreply.github.com>
## Problem
The billing team wants to change the billing events pipeline and use a
common events format in S3 buckets across different event producers.
## Summary of changes
Change the events storage format for billing events from JSON to NDJSON.
Resolves: https://github.com/neondatabase/cloud/issues/29994
## Problem
PGLB will do the connect_to_compute logic, neonkeeper will do the
session establishment logic. We should split it.
## Summary of changes
Moves postgres authentication to compute to a separate routine that
happens after connect_to_compute.
The 1.88.0 stable release is near (this Thursday). We'd like to fix most
warnings beforehand so that the compiler upgrade doesn't require
approval from too many teams.
This is therefore a preparation PR (like similar PRs before it).
There is a lot of changes for this release, mostly because the
`uninlined_format_args` lint has been added to the `style` lint group.
One can read more about the lint
[here](https://rust-lang.github.io/rust-clippy/master/#/uninlined_format_args).
The PR is the result of `cargo +beta clippy --fix` and `cargo fmt`. One
remaining warning is left for the proxy team.
---------
Co-authored-by: Conrad Ludgate <conrad@neon.tech>
## Problem
`Attached(0)` tenant migrations can get stuck if the heatmap file has
not been uploaded.
## Summary of Changes
- Added a test to reproduce the issue.
- Introduced a `kick_secondary_downloads` config flag:
- Enabled in testing environments.
- Disabled in production (and in the new test).
- Updated `Attached(0)` locations to consider the number of secondaries
in their intent when deciding whether to download the heatmap.
## Problem
Part of #11813
## Summary of changes
Add a test API to make it easier to manipulate the feature flags within
tests.
---------
Signed-off-by: Alex Chi Z <chi@neon.tech>
## Problem
The scheduler uses total shards per AZ to select the AZ for newly
created or attached tenants.
This makes bad decisions when we have different node counts per AZ -- we
might have 2 very busy pageservers in one AZ, and 4 more lightly loaded
pageservers in other AZs, and the scheduler picks the busy pageservers
because the total shard count in their AZ is lower.
## Summary of changes
- Divide the shard count by the number of nodes in the AZ when scoring
in `get_az_for_new_tenant`
---------
Co-authored-by: John Spray <john.spray@databricks.com>
The `--gzip-probability` parameter was removed in #12250. However,
`test_basebackup_with_high_slru_count` still uses it, and keeps failing.
This patch removes the use of the parameter (gzip is enabled by
default).
## Problem
Part of #11813
## Summary of changes
The current interval is 30s and it costs a lot of $$$. This patch
reduced it to 600s refresh interval (which means that it takes 10min for
feature flags to propagate from UI to the pageserver). In the future we
can let storcon retrieve the feature flags and push it to pageservers.
We can consider creating a new release or we can postpone this to the
week after the next week.
---------
Signed-off-by: Alex Chi Z <chi@neon.tech>
The 'make all' step must run always. PR #12311 accidentally left the
condition in there to skip it if there were no changes in postgres v14
sources. That condition belonged to a whole different step that was
removed altogether in PR#12311, and the condition should've been removed
too.
Per CI failure:
https://github.com/neondatabase/neon/actions/runs/15820148967/job/44587394469
## Problem
Previously, we couldn't read from an in-memory layer while a batch was
being written to it. Vice-versa, we couldn't write to it while there
was an on-going read.
## Summary of Changes
The goal of this change is to improve concurrency. Writes happened
through a &mut self method so the enforcement was at the type system
level.
We attempt to improve by:
1. Adding interior mutability to EphemeralLayer. This involves wrapping
the buffered writer in a read-write lock.
2. Minimise the time that the read lock is held for. Only hold the read
lock while reading from the buffers (recently flushed or pending
flush). If we need to read from the file, drop the lock and allow IO
to be concurrent.
The new benchmark variants with concurrent reads improve between 70 to
200 percent (against main).
Benchmark results are in this
[commit](891f094ce6).
## Future Changes
We can push the interior mutability into the buffered writer. The
mutable tail goes under a read lock, the flushed part goes into an
ArcSwap and then we can read from anything that is flushed _without_ any
locking.
gRPC base backups send a stream of fixed-size 64KB chunks.
pagebench basebackup with compression enabled shows this to reduce
throughput:
* 64 KB: 55 RPS
* 128 KB: 69 RPS
* 256 KB: 73 RPS
* 1024 KB: 73 RPS
This patch sets the base backup chunk size to 256 KB.
The build-tools image contains various build tools and dependencies,
mostly Rust-related. The compute image build used it to build
compute_ctl and a few other little rust binaries that are included in
the compute image. However, for extensions built in Rust (pgrx), the
build used a different layer which installed the rust toolchain using
rustup.
Switch to using the same rust toolchain for both pgrx-based extensions
and compute_ctl et al. Since we don't need anything else from the
build-tools image, I switched to using the toolchain installed with
rustup, and eliminated the dependency to build-tools altogether. The
compute image build no longer depends on build-tools.
Note: We no longer use 'mold' for linking compute_ctl et al, since mold
is not included in the build-deps-with-cargo layer. We could add it
there, but it doesn't seem worth it. I proposed stopping using mold
altogether in https://github.com/neondatabase/neon/pull/10735, but that
was rejected because 'mold' is faster for incremental builds. That
doesn't matter much for docker builds however, since they're not
incremental, and the compute binaries are not as large as the storage
server binaries anyway.
To avoid duplicating the build logic. `make all` covers the separate
`postgres-*` and `neon-pg-ext` steps, and also does `cargo build`.
That's how you would typically do a full local build anyway.
Every time you run `make`, it runs `make install` on all the PostgreSQL
sources, which copies the header files. That in turn triggers a rebuild
of the `postgres_ffi` crate, and everything that depends on it. We had
worked around this earlier (see #2458), by passing a custom INSTALL
script to the Postgres makefiles, which refrains from updating the
modification timestamp on headers when they have not been changed, but
the v14 makefile didn't obey INSTALL for the header files. Backporting
c0a1d7621b to v14 fixes that.
This backports upstream PostgreSQL commit c0a1d7621b to v14.
Corresponding PR in the 'postgres' repo:
https://github.com/neondatabase/postgres/pull/660
## Problem
In #12217, we began passing the stripe size in reattach responses, and
persisting it in the on-disk state. This is necessary to ensure the
storage controller and Pageserver have a consistent view of the intended
stripe size of unsharded tenants, which will be used for splits that do
not specify a stripe size. However, for backwards compatibility, these
stripe sizes were optional.
## Summary of changes
Make the stripe sizes required for reattach responses and on-disk
location configs. These will always be provided by the previous
(current) release.
The previous behavior was for the compute to override control plane
options if there was a conflict. We want to change the behavior so that
the control plane has the absolute power on what is right. In the event
that we need a new option passed to the compute as soon as possible, we
can initially roll it out in the control plane, and then migrate the
option to EXTRA_OPTIONS within the compute later, for instance.
Signed-off-by: Tristan Partin <tristan@neon.tech>
## Problem
We need to specify the number of safekeepers for neon_local without
`testing` feature.
Also we need this option for testing different configurations of
safekeeper migration code.
We cannot set it in `neon_fixtures.py` and in the default config of
`neon_local` yet, because it will fail compatibility tests. I'll make a
separate PR with removing `cfg!("testing")` completely and specifying
this option in the config when this option reaches the release branch.
- Part of https://github.com/neondatabase/neon/issues/12298
## Summary of changes
- Add `timeline_safekeeper_count` config option to storcon and
neon_local
Enable it across tests and set it as default. Marks the first milestone
of https://github.com/neondatabase/neon/issues/9114. We already enabled
it in all AWS regions and planning to enable it in all Azure regions
next week.
will merge after we roll out in all regions.
---------
Signed-off-by: Alex Chi Z <chi@neon.tech>
## Problem
Timeline GC is very aggressive with regards to layer map locking.
We've seen timelines with loads of layers in production that hold the
write lock for the layer map for 30 minutes at a time.
This blocks reads and the write path to some extent.
## Summary of changes
Determining the set of layers to GC is done under the read lock.
Applying the updates is done under the write lock.
Previously, everything was done under write lock.
See #11942
Idea:
* if connections are short lived, they can get enqueued and then also
remove themselves later if they never made it to redis. This reduces the
load on the queue.
* short lived connections (<10m, most?) will only issue 1 command, we
remove the delete command and rely on ttl.
* we can enqueue as many commands as we want, as we can always cancel
the enqueue, thanks to the ~~intrusive linked lists~~ `BTreeMap`.
This improves `pagebench getpage-latest-lsn` gRPC support by:
* Using `page_api::Client`.
* Removing `--protocol`, and using the `page-server-connstring` scheme
instead.
* Adding `--compression` to enable zstd compression.
2025-06-20 08:31:40 +00:00
540 changed files with 31459 additions and 9150 deletions
# Keep the version the same as in compute/compute-node.Dockerfile
ENVPROTOC_VERSION=25.1
RUN curl -fsSL "https://github.com/protocolbuffers/protobuf/releases/download/v${PROTOC_VERSION}/protoc-${PROTOC_VERSION}-linux-$(uname -m | sed 's/aarch64/aarch_64/g').zip" -o "protoc.zip"\
&& unzip -q protoc.zip -d protoc \
@@ -179,7 +180,7 @@ RUN curl -sL "https://github.com/peak/s5cmd/releases/download/v${S5CMD_VERSION}/
&& mv s5cmd /usr/local/bin/s5cmd
# LLVM
ENVLLVM_VERSION=19
ENVLLVM_VERSION=20
RUN curl -fsSL 'https://apt.llvm.org/llvm-snapshot.gpg.key'| apt-key add - \
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
