Memory,cpu,connections leak with deleted clusters

[andreineustroev]

Describe the bug

When create and delete some amount of clusters (5-10) with apps (~50) we have a memory and cpu leak (up to 8 core, 16Gb). Argo try to connect to deleted clusters and spend a lot of resources on it. It affect reconcile process for alive clusters. For now it's a problem of test environment, where we create and delete a lot of k8s clusters.

Workaround - recreate argo controller pod.

To Reproduce

Create 10 clusters
Delete 10 clusters
Repeat

Expected behavior

Argo delete cluster and free all connected resources correct.

Version

argocd: v2.10.17+6e33cba
  BuildDate: 2024-09-26T06:42:35Z
  GitCommit: 6e33cba80e65f7f60f1a1e8f677a6e2dce315368
  GitTreeState: clean
  GoVersion: go1.21.10
  Compiler: gc
  Platform: linux/amd64

Logs

Nothing special in logs

[rumstead]

@andreineustroev have you tried with a newer version of Argo CD? 2.10 is no longer supported.