Closed bburghaus closed 4 years ago
@bburghaus Can you please provide details on what kind of tolerations do you use? Did you put them on GameServers?
There is a script which actually waits for all pods used by GameServers
to be deleted install/helm/agones/scripts/delete_agones_resources.sh
so it seems that this final step got stuck.
There was an issue with helm delete --purge agones
time out and then helm install ..
gives:
Release "agones" does not exist. Installing it now.
Error: release agones failed: object is being deleted: customresourcedefinitions.apiextensions.k8s.io "gameservers.agones.dev" already exists
make: *** [install] Error 1
In that case patching helps:
$ kubectl patch crd/gameservers.agones.dev -p '{"metadata":{"finalizers":[]}}' --type=merge
Hey @aLekSer, the issue is that all our nodes have taints so the job pod itself cannot be scheduled and the delete hits a timeout.
We could use either --no-hooks
as a solution or agones.crds.cleanupOnDelete=false
, though we want to be able to have the cleanup. We did also remove the finalizers in order to make it work.
The job should take them as separate parameters, or re-use controller tolerations/affinities.
What happened: Helm delete will fail due to the job created by hook script
pre_delete_hook.yaml
if there are tolerations/affinities.What you expected to happen: Successful release deletion when running helm delete.
How to reproduce it (as minimally and precisely as possible): Add tolerations/affinities to cluster, run helm delete.
Environment:
kubectl version
): 1.14