Add a cache for what resource types are scalable so not every reconcile has to do a GET to the /scale endpoint #1453
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
…oint, even when the data isn't needed. Signed-off-by: Noah Kantrowitz <noah@coderanger.net>
|
This did reduce the rate of "Throttling request" warnings but not all of them. |
Signed-off-by: Noah Kantrowitz <noah@coderanger.net>
coderanger
changed the title
Yeah that would definitely help, lack of support of this in Thanks for the fix! |
ycabrer
pushed a commit
to ycabrer/keda
that referenced
this pull request
Mar 1, 2021
…le has to do a GET to the /scale endpoint (kedacore#1453) * Add a cache so not every reconcile has to do a GET to the /scale endpoint, even when the data isn't needed. Signed-off-by: Noah Kantrowitz <noah@coderanger.net> * Add statefulsets to the prefill list too. Signed-off-by: Noah Kantrowitz <noah@coderanger.net>
# for free
to join this conversation on GitHub.
Already have an account?
# 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Outdated suggestions cannot be applied.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.
Adds an in-memory cache to speed up reconciles. With large numbers of scaled objects I'm seeing a ton a request throttling (#1449) that I think (but won't be sure until I try deploying this and seeing if it goes away) is related to these scale requests on every reconcile. The better solution would be for controller-runtime to support access to the scale subresource (kubernetes-sigs/controller-runtime#172) so these requests could hit the local informer cache or something, but that's got a lot of problems itself so I think this solution is a good step for now.
Checklist