Recovery terms in per-queue files instead of DETS #7726
Closed
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
Per-vhost DETS file with recovery terms for all queues is a bottleneck when stopping RabbitMQ - all queues try save their state, leading to a very long file server mailbox and very unpredictable time to stop RabbitMQ (on my machine it can vary from 20 seconds to 5 minutes with 100k classic queues). In this PR we can still read the recovery terms from DETS but we only save them in per-queue files. This way each queue can quickly store its state. Under the same condition, my machine can consistently stop RabbitMQ in 15 seconds or so. The tradeoff is a slower startup time: on my machine, it goes up from 29 seconds to 38 seconds, but that's still better than what we had until #7676 was merged a few days ago. More importantly, the total of stop+start is lower and more predictable. This PR also improves shutdown with many classic queues v1. Startup time with 100k CQv1s is so long and unpredictable that it's hard to even tell if this PR affects it (it varies from 4 to 8 minutes for me). Unfortunately this PR makes startup on MacOS slower (~55s instead of 30s for me), but we don't have to optimise for that. In most cases (with much fewer queues), it won't be noticable anyway.
|
Unfortunately the startup time trade-off of this approach is very high in my GKE tests: 52s for |
|
A completely different approach is in the works. |
This was referenced Mar 24, 2023
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Outdated suggestions cannot be applied.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.
Note: this PR maintains backwards compatibility; a separate PR will remove the old bits
Per-vhost DETS file with recovery terms for all queues is a bottleneck when stopping RabbitMQ - all queues try save their state, leading to a very long file server mailbox and very unpredictable time to stop RabbitMQ (on my machine it can vary from 20 seconds to 5 minutes with 100k classic queues).
In this PR we can still read the recovery terms from DETS but we only save them in per-queue files. This way each queue can quickly store its state. Under the same condition, my machine can consistently stop RabbitMQ in 15 seconds or so.
The tradeoff is a slower startup time: on my machine, it goes up from 29 seconds to 38 seconds, but that's still better than what we had until #7676 was merged a few days ago. More importantly, the total of stop+start is lower and more predictable.
This PR also improves shutdown with many classic queues v1. Startup time with 100k CQv1s is so long and unpredictable that it's hard to even tell if this PR affects it (it varies from 4 to 8 minutes for me).
Unfortunately this PR makes startup on MacOS slower (~55s instead of 30s for me), but we don't have to optimise for that. In most cases (with much fewer queues), it won't be noticeable anyway.