-
Notifications
You must be signed in to change notification settings - Fork 5.6k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
argocd-repo-server OOMKilled after Kubernetes version 1.30 upgrade #19740
Comments
Can you confirm that this issue is reproduced in the latest version of argocd? |
can u try argocd 2.12.x ? |
@Jack-R-lantern @tooptoop4 Unfortunately, we cannot upgrade to more recent argocd version due to some blockers. Is there any possible reasons you see that may have caused argocd-repo-server to be OOMKilled specifically after Kubernetes 1.30 upgrade? We have not experienced similar issues in previous Kubernetes upgrades and previous cases where all argocd pods were restarted. |
I know others have said this but version See the testing matrix for supported versions https://argo-cd.readthedocs.io/en/stable/operator-manual/tested-kubernetes-versions/ |
My guess is that many API calls to request the repo content with a cache miss at the same time. You can try to set the value to be smaller number to see if it helps, but it will also limit the performance of repo server. |
To pile on, compression has also been enabled by default since v2.8. |
@christianh814 @wanghong230 @rumstead Thank you for all the insights! Much appreciated 🙏 Theoretically, enabling redis gzip compression even in 2.7.14 should help with |
Checklist:
argocd version
.Describe the bug
Recently, we had a case where all argocd components were reassigned to nodes after a kubernetes upgrade. After the upgrade, argocd-repo-server pod was seen to be constantly getting OOMKilled. When we check metrics, there is no sign of constant high memory usage, so it seems like there are some memory usage peaking. We are struggling to find the cause of this memory usage peaking, and why particularly after the Kubernetes upgrade.
More info:
/tmp
directory forargocd-repo-server
container and cmp sidecar container respectively.Is there any reason why argocd-repo-server may have memory usage peaking after all argocd components were restarted?
Any insights would be appreciated!
To Reproduce
Expected behavior
argocd-repo-server pod's main container and cmp sidecar container not getting OOMKilled (no memory usage peaking)
Screenshots
Version
Logs
logs from argocd-repo-server container before it got OOMKilled
The text was updated successfully, but these errors were encountered: