cherrypick-1.1: distsql: pre-reserve memory needed to mark rows in HashJoiner build phase #18975
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
…hase A situation was uncovered by #18600, where the HashJoiner would run out of memory in the probe phase. This was because we had made an assumption that we wouldn't hit a memory limit if the buffer phase filled up at most 2/3 of the limit with both streams , since the marking infrastructure would take up only a fraction of 1/3 (the chosen stream). This assumption failed to take into account other limits shared with other processors. This change pre-reserves the memory needed for the probe phase in the build phase so that we can keep a single point in the code where we fall back to disk while not relying on any limit assumptions.
|
This change is |
|
Review status: 0 of 3 files reviewed at latest revision, all discussions resolved, all commit checks successful. Comments from Reviewable |
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Outdated suggestions cannot be applied.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.
A situation was uncovered by #18600, where the HashJoiner would run out
of memory in the probe phase. This was because we had made an assumption
that we wouldn't hit a memory limit if the buffer phase filled up at
most 2/3 of the limit with both streams , since the marking
infrastructure would take up only a fraction of 1/3 (the chosen stream).
This assumption failed to take into account other limits shared with
other processors. This change pre-reserves the memory needed for the
probe phase in the build phase so that we can keep a single point in the
code where we fall back to disk while not relying on any limit
assumptions.