page_service: batching observability & include throttled time in smgr metrics #9870
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
The steps in the test work in neon_local + psql but for some reason they don't work in the test. Asked compute team on Slack for help: https://neondb.slack.com/archives/C04DGM6SMTM/p1731952688386789
This reverts commit b974616.
This reverts commit aa695b2.
With this, 10us batching timeout works, but it has some other wrinkles: - it uses the signal-based timer APIs instead of going through epoll (=> timerfd) = it needs to make a syscall for each batch, which costs around 1-2us, so, probably significant CPU time wasted on this.
This reverts commit 1639b26.
batching at 10us doesn't work well enough, prob the future is ready too soon. batching factor is just 1.5 https://www.notion.so/neondatabase/benchmarking-notes-143f189e004780c4a630cb5f426e39ba?pvs=4#144f189e004780b79c8dd6d007dbb120
This reverts commit 81d9970.
Resolution not high enough to do _any_ batching at 10us or 20us https://www.notion.so/neondatabase/benchmarking-notes-143f189e004780c4a630cb5f426e39ba?pvs=4#144f189e0047800fb74bd8f4ab6cf8e2
This reverts commit 12124b2.
Yep, it's clearly the best one with best batching factor at lowest CPU usage. https://www.notion.so/neondatabase/benchmarking-notes-143f189e004780c4a630cb5f426e39ba?pvs=4#144f189e004780d0a205e081458b46db
Best batching factor so far with no worse degradation of un-batchable workloads than the other candidates. https://www.notion.so/neondatabase/benchmarking-notes-143f189e004780c4a630cb5f426e39ba?pvs=4#144f189e004780c0921fe99e1da0e8c9
This reverts commit 721643b.
This reverts commit 68550f0.
This reverts commit c73e9e4.
This reverts commit 689788c.
Performs identically great to the async-timer::Timer features=tokio1 impl Makes sense because it's the same thing that's happening under the hood. https://www.notion.so/neondatabase/benchmarking-notes-143f189e004780c4a630cb5f426e39ba?pvs=4#144f189e004780ea9decc82281f6b8d1
problame
changed the base branch from
problame/batching-sidecar-task
to
main
problame
force-pushed
the
problame/batching-metrics-improvements
branch
2 times, most recently
from
9193a4f to
d04f05d
Compare
…rics-improvements
problame
force-pushed
the
problame/batching-metrics-improvements
branch
from
d04f05d to
69b878f
Compare
|
Opened a discussion in https://neondb.slack.com/archives/C033RQ5SPDH/p1732910861704029 |
|
Potential solution: #9962 |
VladLazar
reviewed
Dec 2, 2024
VladLazar
approved these changes
Dec 2, 2024
github-merge-queue bot
pushed a commit
that referenced
this pull request
Dec 3, 2024
… deduction for smgr latency metrics (#9962) ## Problem In the batching PR - #9870 I stopped deducting the time-spent-in-throttle fro latency metrics, i.e., - smgr latency metrics (`SmgrOpTimer`) - basebackup latency (+scan latency, which I think is part of basebackup). The reason for stopping the deduction was that with the introduction of batching, the trick with tracking time-spent-in-throttle inside RequestContext and swap-replacing it from the `impl Drop for SmgrOpTimer` no longer worked with >1 requests in a batch. However, deducting time-spent-in-throttle is desirable because our internal latency SLO definition does not account for throttling. ## Summary of changes - Redefine throttling to be a page_service pagestream request throttle instead of a throttle for repository `Key` reads through `Timeline::get` / `Timeline::get_vectored`. - This means reads done by `basebackup` are no longer subject to any throttle. - The throttle applies after batching, before handling of the request. - Drive-by fix: make throttle sensitive to cancellation. - Rename metric label `kind` from `timeline_get` to `pagestream` to reflect the new scope of throttling. To avoid config format breakage, we leave the config field named `timeline_get_throttle` and ignore the `task_kinds` field. This will be cleaned up in a future PR. ## Trade-Offs Ideally, we would apply the throttle before reading a request off the connection, so that we queue the minimal amount of work inside the process. However, that's not possible because we need to do shard routing. The redefinition of the throttle to limit pagestream request rate instead of repository `Key` rate comes with several downsides: - We're no longer able to use the throttle mechanism for other other tasks, e.g. image layer creation. However, in practice, we never used that capability anyways. - We no longer throttle basebackup.
awarus
pushed a commit
that referenced
this pull request
Dec 5, 2024
… metrics (#9870) This PR - fixes smgr metrics #9925 - adds an additional startup log line logging the current batching config - adds a histogram of batch sizes global and per-tenant - adds a metric exposing the current batching config The issue described #9925 is that before this PR, request latency was only observed *after* batching. This means that smgr latency metrics (most importantly getpage latency) don't account for - `wait_lsn` time - time spent waiting for batch to fill up / the executor stage to pick up the batch. The fix is to use a per-request batching timer, like we did before the initial batching PR. We funnel those timers through the entire request lifecycle. I noticed that even before the initial batching changes, we weren't accounting for the time spent writing & flushing the response to the wire. This PR drive-by fixes that deficiency by dropping the timers at the very end of processing the batch, i.e., after the `pgb.flush()` call. I was **unable to maintain the behavior that we deduct time-spent-in-throttle from various latency metrics. The reason is that we're using a *single* counter in `RequestContext` to track micros spent in throttle. But there are *N* metrics timers in the batch, one per request. As a consequence, the practice of consuming the counter in the drop handler of each timer no longer works because all but the first timer will encounter error `close() called on closed state`. A failed attempt to maintain the current behavior can be found in #9951. So, this PR remvoes the deduction behavior from all metrics. I started a discussion on Slack about it the implications this has for our internal SLO calculation: https://neondb.slack.com/archives/C033RQ5SPDH/p1732910861704029 # Refs - fixes #9925 - sub-issue #9377 - epic: #9376
awarus
pushed a commit
that referenced
this pull request
Dec 5, 2024
… deduction for smgr latency metrics (#9962) ## Problem In the batching PR - #9870 I stopped deducting the time-spent-in-throttle fro latency metrics, i.e., - smgr latency metrics (`SmgrOpTimer`) - basebackup latency (+scan latency, which I think is part of basebackup). The reason for stopping the deduction was that with the introduction of batching, the trick with tracking time-spent-in-throttle inside RequestContext and swap-replacing it from the `impl Drop for SmgrOpTimer` no longer worked with >1 requests in a batch. However, deducting time-spent-in-throttle is desirable because our internal latency SLO definition does not account for throttling. ## Summary of changes - Redefine throttling to be a page_service pagestream request throttle instead of a throttle for repository `Key` reads through `Timeline::get` / `Timeline::get_vectored`. - This means reads done by `basebackup` are no longer subject to any throttle. - The throttle applies after batching, before handling of the request. - Drive-by fix: make throttle sensitive to cancellation. - Rename metric label `kind` from `timeline_get` to `pagestream` to reflect the new scope of throttling. To avoid config format breakage, we leave the config field named `timeline_get_throttle` and ignore the `task_kinds` field. This will be cleaned up in a future PR. ## Trade-Offs Ideally, we would apply the throttle before reading a request off the connection, so that we queue the minimal amount of work inside the process. However, that's not possible because we need to do shard routing. The redefinition of the throttle to limit pagestream request rate instead of repository `Key` rate comes with several downsides: - We're no longer able to use the throttle mechanism for other other tasks, e.g. image layer creation. However, in practice, we never used that capability anyways. - We no longer throttle basebackup.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR
pageserver_smgr_query_started_countincremented after waiting for effective lsn #9925The issue described #9925 is that before this PR, request latency was only observed after batching.
This means that smgr latency metrics (most importantly getpage latency) don't account for
wait_lsntimeThe fix is to use a per-request batching timer, like we did before the initial batching PR.
We funnel those timers through the entire request lifecycle.
I noticed that even before the initial batching changes, we weren't accounting for the time spent writing & flushing the response to the wire.
This PR drive-by fixes that deficiency by dropping the timers at the very end of processing the batch, i.e., after the
pgb.flush()call.I was **unable to maintain the behavior that we deduct time-spent-in-throttle from various latency metrics.
The reason is that we're using a single counter in
RequestContextto track micros spent in throttle.But there are N metrics timers in the batch, one per request.
As a consequence, the practice of consuming the counter in the drop handler of each timer no longer works because all but the first timer will encounter error
close() called on closed state.A failed attempt to maintain the current behavior can be found in #9951.
So, this PR remvoes the deduction behavior from all metrics.
I started a discussion on Slack about it the implications this has for our internal SLO calculation: https://neondb.slack.com/archives/C033RQ5SPDH/p1732910861704029
Refs
pageserver_smgr_query_started_countincremented after waiting for effective lsn #9925