v1.2: OOMKilled of openshift-repo-server after upgrade from 1.1.2 about gitops-operator HOT 5 CLOSED

Increasing the limit to 1Gi was sufficient for my setup.

from gitops-operator.

@EagleIJoe Thanks for confirmation. May I ask for details about your setup? I.e. do you use a mono repo? How many apps? What kind of toolings? This information could help us in fine-tuning resource requests in future releases, or at least provide some more guidance to users.

from gitops-operator.

@EagleIJoe On which cluster it is reported ? Screenshot and failure log would be helpful.

from gitops-operator.

Hey, we do apply some resource limits by default now which might be too low for some larger installations or when using custom plugins.

You can manually alter the limits in the ArgoCD CR. An example to increase memory limit for the repo-server:

spec:
  repo:
    resources:
       limits:
         memory: <some value>

The default limit is 512Mi which should be sufficient for most installations. You can also try and scale out the repo server workload by increasing its numbers of replicas to distribute workload across more pods.

from gitops-operator.

@jannfis @iam-veeramalla i have the following setup:

• OpenShift 4.7
• 100 Apps deployed via appsets pulling from a monorepo
• rendered via kustomize
• 1 argocd repo server instance, 1 argocd server instance

So nothing major. What I am seeing now is the infamous "all SubConns are in TransientFailure" errors happening...

This was working fine in GitOps 1.1.2.

from gitops-operator.