Bypass MMap arena grouping as this has caused issues with too many regions being mapped #135012
+43
−5
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
…gions being mapped
benwtrent
added
>bug
:Core/Infra/Core
|
Hi @benwtrent, I've created a changelog YAML for you. |
…rent/elasticsearch into bugfix/bypass-mmmap-arena-grouping
elasticsearchmachine
added
Team:Core/Infra
|
Pinging @elastic/es-search (Team:Search) |
|
Pinging @elastic/es-core-infra (Team:Core/Infra) |
|
Pinging @elastic/es-distributed-indexing (Team:Distributed Indexing) |
|
I don't have anything to say about the code but i agree with the idea to disable this logic given the amount of issues we've had. |
ChrisHegarty
approved these changes
Sep 19, 2025
There was a problem hiding this comment.
@benwtrent and I chatted abut this offline, and the changes look good to me.
benwtrent
added
auto-merge-without-approval
💔 Backport failed
You can use sqren/backport to manually backport by running |
benwtrent
added a commit
to benwtrent/elasticsearch
that referenced
this pull request
Sep 19, 2025
…gions being mapped (elastic#135012) There is a JDK issue where closing sharedArenas from many threads can significantly harm performance. This ref-counting of shared arenas was designed as a way to get around this performance issue. However, we have noticed a significant increase in leaks and issues with mmap regions since this change. https://bugs.openjdk.org/browse/JDK-8335480 should have helped the performance impact of closing shared arenas (though possibly not fully mitigated it). I am proposing we turn off the grouping as it appears (at least to me), not worth it. I am willing to backdown if we thing other fixes should be done. I also suggest this gets backported to 9.1, 8.19, and is merged into 9.2
elasticsearchmachine
pushed a commit
that referenced
this pull request
Sep 19, 2025
…gions being mapped (#135012) (#135115) There is a JDK issue where closing sharedArenas from many threads can significantly harm performance. This ref-counting of shared arenas was designed as a way to get around this performance issue. However, we have noticed a significant increase in leaks and issues with mmap regions since this change. https://bugs.openjdk.org/browse/JDK-8335480 should have helped the performance impact of closing shared arenas (though possibly not fully mitigated it). I am proposing we turn off the grouping as it appears (at least to me), not worth it. I am willing to backdown if we thing other fixes should be done. I also suggest this gets backported to 9.1, 8.19, and is merged into 9.2
💚 All backports created successfully
Questions ?Please refer to the Backport tool documentation |
💚 All backports created successfully
Questions ?Please refer to the Backport tool documentation |
benwtrent
added a commit
to benwtrent/elasticsearch
that referenced
this pull request
Sep 20, 2025
…gions being mapped (elastic#135012) There is a JDK issue where closing sharedArenas from many threads can significantly harm performance. This ref-counting of shared arenas was designed as a way to get around this performance issue. However, we have noticed a significant increase in leaks and issues with mmap regions since this change. https://bugs.openjdk.org/browse/JDK-8335480 should have helped the performance impact of closing shared arenas (though possibly not fully mitigated it). I am proposing we turn off the grouping as it appears (at least to me), not worth it. I am willing to backdown if we thing other fixes should be done. I also suggest this gets backported to 9.1, 8.19, and is merged into 9.2 (cherry picked from commit 2672cd0)
benwtrent
added a commit
to benwtrent/elasticsearch
that referenced
this pull request
Sep 20, 2025
…gions being mapped (elastic#135012) There is a JDK issue where closing sharedArenas from many threads can significantly harm performance. This ref-counting of shared arenas was designed as a way to get around this performance issue. However, we have noticed a significant increase in leaks and issues with mmap regions since this change. https://bugs.openjdk.org/browse/JDK-8335480 should have helped the performance impact of closing shared arenas (though possibly not fully mitigated it). I am proposing we turn off the grouping as it appears (at least to me), not worth it. I am willing to backdown if we thing other fixes should be done. I also suggest this gets backported to 9.1, 8.19, and is merged into 9.2 (cherry picked from commit 2672cd0)
elasticsearchmachine
pushed a commit
that referenced
this pull request
Sep 20, 2025
…any regions being mapped (#135012) (#135131) * Bypass MMap arena grouping as this has caused issues with too many regions being mapped (#135012) There is a JDK issue where closing sharedArenas from many threads can significantly harm performance. This ref-counting of shared arenas was designed as a way to get around this performance issue. However, we have noticed a significant increase in leaks and issues with mmap regions since this change. https://bugs.openjdk.org/browse/JDK-8335480 should have helped the performance impact of closing shared arenas (though possibly not fully mitigated it). I am proposing we turn off the grouping as it appears (at least to me), not worth it. I am willing to backdown if we thing other fixes should be done. I also suggest this gets backported to 9.1, 8.19, and is merged into 9.2 (cherry picked from commit 2672cd0) * [CI] Auto commit changes from spotless * fixing format * fixing compilation --------- Co-authored-by: elasticsearchmachine <[email protected]>
elasticsearchmachine
pushed a commit
that referenced
this pull request
Sep 20, 2025
…many regions being mapped (#135012) (#135133) * Bypass MMap arena grouping as this has caused issues with too many regions being mapped (#135012) There is a JDK issue where closing sharedArenas from many threads can significantly harm performance. This ref-counting of shared arenas was designed as a way to get around this performance issue. However, we have noticed a significant increase in leaks and issues with mmap regions since this change. https://bugs.openjdk.org/browse/JDK-8335480 should have helped the performance impact of closing shared arenas (though possibly not fully mitigated it). I am proposing we turn off the grouping as it appears (at least to me), not worth it. I am willing to backdown if we thing other fixes should be done. I also suggest this gets backported to 9.1, 8.19, and is merged into 9.2 (cherry picked from commit 2672cd0) * [CI] Auto commit changes from spotless * fixing compilation --------- Co-authored-by: elasticsearchmachine <[email protected]>
elasticsearchmachine
pushed a commit
that referenced
this pull request
Sep 20, 2025
…many regions being mapped (#135012) (#135130) * Bypass MMap arena grouping as this has caused issues with too many regions being mapped (#135012) There is a JDK issue where closing sharedArenas from many threads can significantly harm performance. This ref-counting of shared arenas was designed as a way to get around this performance issue. However, we have noticed a significant increase in leaks and issues with mmap regions since this change. https://bugs.openjdk.org/browse/JDK-8335480 should have helped the performance impact of closing shared arenas (though possibly not fully mitigated it). I am proposing we turn off the grouping as it appears (at least to me), not worth it. I am willing to backdown if we thing other fixes should be done. I also suggest this gets backported to 9.1, 8.19, and is merged into 9.2 (cherry picked from commit 2672cd0) * [CI] Auto commit changes from spotless * fixing compilation --------- Co-authored-by: elasticsearchmachine <[email protected]>
gmjehovich
pushed a commit
to gmjehovich/elasticsearch
that referenced
this pull request
Sep 22, 2025
…gions being mapped (elastic#135012) There is a JDK issue where closing sharedArenas from many threads can significantly harm performance. This ref-counting of shared arenas was designed as a way to get around this performance issue. However, we have noticed a significant increase in leaks and issues with mmap regions since this change. https://bugs.openjdk.org/browse/JDK-8335480 should have helped the performance impact of closing shared arenas (though possibly not fully mitigated it). I am proposing we turn off the grouping as it appears (at least to me), not worth it. I am willing to backdown if we thing other fixes should be done. I also suggest this gets backported to 9.1, 8.19, and is merged into 9.2
DonalEvans
pushed a commit
to DonalEvans/elasticsearch
that referenced
this pull request
Sep 22, 2025
…gions being mapped (elastic#135012) There is a JDK issue where closing sharedArenas from many threads can significantly harm performance. This ref-counting of shared arenas was designed as a way to get around this performance issue. However, we have noticed a significant increase in leaks and issues with mmap regions since this change. https://bugs.openjdk.org/browse/JDK-8335480 should have helped the performance impact of closing shared arenas (though possibly not fully mitigated it). I am proposing we turn off the grouping as it appears (at least to me), not worth it. I am willing to backdown if we thing other fixes should be done. I also suggest this gets backported to 9.1, 8.19, and is merged into 9.2
|
++ from me as well, thanks for opening this. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
6 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.
There is a JDK issue where closing sharedArenas from many threads can significantly harm performance.
This ref-counting of shared arenas was designed as a way to get around this performance issue. However, we have noticed a significant increase in leaks and issues with mmap regions since this change.
https://bugs.openjdk.org/browse/JDK-8335480 should have helped the performance impact of closing shared arenas (though possibly not fully mitigated it).
I am proposing we turn off the grouping as it appears (at least to me), not worth it.
I am willing to backdown if we thing other fixes should be done.
I also suggest this gets backported to 9.1, 8.19, and is merged into 9.2