fix: honor stacktrace partitions at downsampling #3408
Merged
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
kolesnikovae
force-pushed
the
fix/downsample-by-partition
branch
from
1207496
to
31da2f2
Compare
simonswine
approved these changes
Jul 8, 2024
Yeah, I'll cherry pick it |
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Outdated suggestions cannot be applied.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.
Fixes #3406.
Currently, during compaction, when the downsampler aggregates profiles, the stack trace partition is ignored. The initial assumption was that profiles of the same series (label set) would always have the same stack trace partition. This is almost always the case; however, this is not guaranteed as the partitioning heuristic also depends on the main mapping name (see #3262).
We've recently discovered a few cases where this led to data inconsistencies. Specifically, profiles that refer to the wrong partition, and therefore can't be read: their samples refer to stack traces that can't be found in the partition. Another issue is that such occurrences are not handled properly: a wrong stack trace may be resolved to wrong locations (the stack trace buffer is not cleared if the stack trace is not found), which will likely cause a panic when attempting to resolve it.
I also increased the symbol rewriter LRU cache size from 2 to 8, as I've seen that the interlacing of partitions might be quite heavy.