[native]Fix duplicate root memory pool in case of background arbitration#21385
Merged
xiaoxmeng merged 1 commit intoprestodb:masterfrom Nov 15, 2023
Merged
[native]Fix duplicate root memory pool in case of background arbitration#21385xiaoxmeng merged 1 commit intoprestodb:masterfrom
xiaoxmeng merged 1 commit intoprestodb:masterfrom
Conversation
mbasmanova
approved these changes
Nov 14, 2023
Contributor
mbasmanova
left a comment
mbasmanova
left a comment
There was a problem hiding this comment.
Thank you for the fix.
If a task from a query completes before the other tasks get scheduled on the same worker, then the query context might have been destroyed before creating the other tasks from the same query. The query context manager maintains a query context cache but it only holds the weak pointer. So that when the next task from the same query comes, it will create a new query context with the same query id and creates a root memory pool with the same name from the memory manager. If there is a background memory arbitration process, then the old memory pool instance might still be held by the memory arbitration process and run into duplicate root memory pool issue. This PR fixes this issue by appending a monotonically increasing id to root memory pool name to ensure the instance name is unique.
26 tasks
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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 a task from a query completes before the other tasks get scheduled on the same
worker, then the query context might have been destroyed before creating the other
tasks from the same query. The query context manager maintains a query context
cache but it only holds the weak pointer. So that when the next task from the same
query comes, it will create a new query context with the same query id and creates
a root memory pool with the same name from the memory manager. If there is a
background memory arbitration process, then the old memory pool instance might
still be held by the memory arbitration process and run into duplicate root memory pool
issue.
This PR fixes this issue by appending a monotonically increasing id to root memory
pool name to ensure the instance name is unique.