storage: tick only raft groups which are not quiesced #26910
Merged
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
Previously we were ticking every replica, regardless of whether or not it was part of a quiesced range, at each tick interval. For 10,000 ranges this was noticeable. Each `Replica.tick()` operation enqueues on a scheduler queue, is serviced by a goroutine which calls `time.Now()`, and acquires several mutexes before discovering it's quiesced and exiting. We happen to have a carefully maintained map of unquiesced replicas, which this change uses instead of all of the store's replicas, when populating the `rangeIDs` slice used to enqueue Raft ticks. Release note (performance improvement): less CPU utilization with many ranges.
|
This change is |
|
FYI, @bdarnell previously made this change (he introduced the Review status: Comments from Reviewable |
complete! 0 of 0 LGTMs obtained
|
Yes, this reverts #26291. I didn't find any other bugs in the use of this map so it's time to give this another shot. Review status: Comments from Reviewable |
nvanbenschoten
approved these changes
Jun 26, 2018
|
Ok, I'll merge this for Spencer then while he's out. bors r=bdarnell |
craig bot
pushed a commit
that referenced
this pull request
Jun 26, 2018
26910: storage: tick only raft groups which are not quiesced r=bdarnell a=spencerkimball Previously we were ticking every replica, regardless of whether or not it was part of a quiesced range, at each tick interval. For 10,000 ranges this was noticeable. Each `Replica.tick()` operation enqueues on a scheduler queue, is serviced by a goroutine which calls `time.Now()`, and acquires several mutexes before discovering it's quiesced and exiting. We happen to have a carefully maintained map of unquiesced replicas, which this change uses instead of all of the store's replicas, when populating the `rangeIDs` slice used to enqueue Raft ticks. Release note (performance improvement): less CPU utilization with many ranges. Co-authored-by: Spencer Kimball <[email protected]>
Build succeeded |
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.
Previously we were ticking every replica, regardless of whether
or not it was part of a quiesced range, at each tick interval.
For 10,000 ranges this was noticeable. Each
Replica.tick()operation enqueues on a scheduler queue, is serviced by a
goroutine which calls
time.Now(), and acquires several mutexesbefore discovering it's quiesced and exiting.
We happen to have a carefully maintained map of unquiesced
replicas, which this change uses instead of all of the store's
replicas, when populating the
rangeIDsslice used to enqueueRaft ticks.
Release note (performance improvement): less CPU utilization
with many ranges.