Fix memory accounting in case of MEMORY_LIMIT_EXCEEDED errors #40249
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
robot-ch-test-poll
added
the
pr-improvement
nikitamikhaylov
assigned nikitamikhaylov and KochetovNicolai and unassigned nikitamikhaylov
KochetovNicolai
approved these changes
Aug 16, 2022
azat
commented
Aug 16, 2022
In case of memory allocation had been failed, it should not be take into account anywhere, otherwise it will report incorrect memory usage for query and in logs. I found this while was digging into OOM in stress tests, and saw memory allocations that had been significantly greater the memory limit (even with overcommit). Signed-off-by: Azat Khuzhin <a.khuzhin@semrush.com>
azat
force-pushed
the
fix-memory-account-on-error
branch
from
0673f92
to
8e4c967
Compare
|
ASTFuzzer: #39991 |
KochetovNicolai
added
the
skip mergeable check
Yeah, I was on it actually (that's why I marked it as draft). |
This was referenced Aug 17, 2022
azat
commented
Aug 20, 2022
| /// more. It could be useful to enlarge Exception message in rethrow logic. | ||
| Int64 tmp = current_thread->untracked_memory; | ||
| current_thread->untracked_memory_limit_increase = current_thread->untracked_memory_limit; |
There was a problem hiding this comment.
@KochetovNicolai found and interesting case here - #40424 (comment)
There was a problem hiding this comment.
But I'm suggesting #40321 anyway
azat
added a commit
to azat/ClickHouse
that referenced
this pull request
Aug 22, 2022
It looks useless nowadays, because operator new cannot throw MEMORY_LIMIT_EXCEEDED today, and so any code that works on Exception is likely safe. Refs: ClickHouse#40249 Refs: ClickHouse#24483 Signed-off-by: Azat Khuzhin <a.khuzhin@semrush.com>
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.
Changelog category (leave one):
Changelog entry (a user-readable short description of the changes that goes to CHANGELOG.md):
Fix memory accounting in case of MEMORY_LIMIT_EXCEEDED errors (previously [peak] memory usage was takes failed allocations into account)
In case of memory allocation had been failed, it should not be take into
account anywhere, otherwise it will report incorrect memory usage for
query and in logs.
I found this while was digging into OOM in stress tests, and saw memory
allocations that had been significantly greater the memory limit (even
with overcommit), where something interesting happened,
consider the query with failed allocation for 1TiB,
but since
ThreadStatus::untracked_memorystill includes it,it will be "freed" in ThreadStatus dtor, however it will not free but allocate,
since
untracked_memory > 0, and since the code called from dtor,exceptions from MemoryTracker will be locked, and amount/peak will be updated.