Faster+Simpler CanMatchPreFilterSearchPhase #116881
Closed
Conversation
This file contains hidden or 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
This redose CanMatchPreFilterSearchPhase to be simpler and without the needless heavy locking. We were locking on every shard result which could introduce a lot of contention when a large number of data nodes (especially when they hold large shard counts) are queried. We probably could even do without the volatile in some cases but either way, this commit does away with any contention on the results since we will never access the same cell in the result array from more than a single thread. There is no need to have a bitset here and deal with the synchronization it entails when we at the same time also set up two arrays of shard_count length and only need a single array to begin with. => share array for failures, min-max and matches, remove all synchronization, further simplify the code.
original-brownbear
added
>non-issue
:Search Foundations/Search
Collaborator
|
Pinging @elastic/es-search-foundations (Team:Search Foundations) |
elasticsearchmachine
added
Team:Search Foundations
original-brownbear
added a commit
that referenced
this pull request
Jan 9, 2025
We don't need the result to be a `SearchPhaseResults`. In fact, there is no reason for it to be a nested class in the first place. Just flatten it into the phase itself and synchronize on `this`. This is just a step on the way to #116881 that makes that PR much easier to review I believe.
elasticsearchmachine
pushed a commit
that referenced
this pull request
Mar 14, 2025
We don't need the result to be a `SearchPhaseResults`. In fact, there is no reason for it to be a nested class in the first place. Just flatten it into the phase itself and synchronize on `this`. This is just a step on the way to #116881 that makes that PR much easier to review I believe.
Contributor
Author
|
This can be closed, no need to invest more in can_match when #121885 is incoming. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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.
If this is too many changes in one PR, happy to split this up. We should fix this though, it's the reason why we're seeing slow warn logs for can_match responses in logs.
This redoes
CanMatchPreFilterSearchPhaseto be simpler and without the needless heavy locking.We were locking on every shard result which could introduce a lot of contention (on the transport threads!) when a large number of data nodes (especially when they hold large shard counts) are queried. We probably could even do without the volatile in some cases but either way, this commit does away with any contention on the results since we will never access the same cell in the result array from more than a single thread.
There is no need to have a bitset here and deal with the synchronization it entails when we at the same time also set up two arrays of shard_count length and only need a single array to begin with.
=> share array for failures, min-max and matches, remove all synchronization, further simplify the code because we don't need the results object if all results just go to a single array :).