Skip to content

Conversation

@shuangkun
Copy link
Member

@shuangkun shuangkun commented Oct 20, 2025

Create new cache instead of writeInformerBack.

Fixes #14833
Fixes #14780
Fixes #12352

Motivation

Modifications

Verification

Documentation

shuangkun and others added 4 commits October 20, 2025 20:10
Signed-off-by: shuangkun <[email protected]>
Signed-off-by: shuangkun <[email protected]>
Signed-off-by: shuangkun <[email protected]>
@eduardodbr
Copy link
Member

This PR is very similar to #14532 , both are adding a new cache in front of the informar cache, which might feel weird. Please review and join the discussion if you have a different opinion

Signed-off-by: shuangkun <[email protected]>
Signed-off-by: shuangkun <[email protected]>
Signed-off-by: shuangkun <[email protected]>
@shuangkun
Copy link
Member Author

This PR is very similar to #14532 , both are adding a new cache in front of the informar cache, which might feel weird. Please review and join the discussion if you have a different opinion

Oh, the problems we're trying to solve may be somewhat different, but there's also some overlap, such as double pod creation. I'm trying to address controller informer anomalies at very large scales (300,000+ workflows), such as event overwriting and workflow controllers missing workflows. Therefore, I'd like to remove the "writeInformerBack" method, which is the official recommended approach: instead of writing to the same InformerCache, create new expectations.

Signed-off-by: shuangkun <[email protected]>
Signed-off-by: shuangkun <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Pod Created twice when cluster under pressure WriteBack Informer not Take effect Add some doc about environment variable "INFORMER_WRITE_BACK"

2 participants