Share data loader to across asyncio boto sessions #40658
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
By default, a botocore session creates and caches an instance of JSONDecoder which consumes a lot of memory. This issue was reported here boto/botocore#3078. In the context of triggers which use boto sessions, this can result in excessive memory usage and as a result reduced capacity on the triggerer. We can reduce memory footprint by sharing the loader instance across the sessions.
boring-cyborg
bot
added
area:providers
provider:amazon-aws
potiuk
approved these changes
Jul 8, 2024
This was referenced Jul 9, 2024
romsharon98
pushed a commit
to romsharon98/airflow
that referenced
this pull request
Jul 26, 2024
By default, a botocore session creates and caches an instance of JSONDecoder which consumes a lot of memory. This issue was reported here boto/botocore#3078. In the context of triggers which use boto sessions, this can result in excessive memory usage and as a result reduced capacity on the triggerer. We can reduce memory footprint by sharing the loader instance across the sessions.
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Outdated suggestions cannot be applied.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.
By default, a botocore session creates and caches an instance of JSONDecoder which consumes a lot of memory. This issue was reported here boto/botocore#3078. In the context of triggers which use boto sessions, this can result in excessive memory usage and as a result reduced capacity on the triggerer. We can reduce memory footprint by sharing the loader instance across the sessions.
Idea for the solution here comes from this comment:
boto/botocore#3078 (comment)
Def let me know if you see any concerns with this.
Below here is memory utilization shown before and after the fix, with 50 s3 key triggers running:
Before (770MB):
After (370MB):
