I run helm ls --all and get
NAME REVISION UPDATED STATUS CHART NAMESPACE
assets 1 Fri Aug 18 15:34:12 2017 DELETED elasticsearch-0.1.9 defaultThen I run helm del --purge assets and it hangs indefinitely.
The tiller pod log moves when I run the ls command, but doesn't move when I run the del command.
I tried deleting the tiller pod to get a fresh one, but the problem remained.
This is the script I used to deploy Elasticsearch in case it helps:
echo "deploying Elasticsearch cluster"
helm init
helm repo add incubator http://storage.googleapis.com/kubernetes-charts-incubator
helm install incubator/elasticsearch --name assets --set data.resources.limits.cpu=250m,data.resources.limits.memory=1024Mi,data.resources.requests.cpu=250m,data.resources.requests.memory=1024MiAny ideas?
After a few tries, I finally got some feedback from the tiller logs:
uninstall: Failed deletion of "assets": timed out waiting for "assets-elasticsearch-data" to be syncedAfter a very long wait and a few restarts of the tiller pod, the delete command worked. So the answer to my initial question is:
Helm can't delete a deployment when not all of its components are synced.
As for why a component was not synced, it's still not clear.
If you have any post install hook that is causing this failure, for helm v3 you can use
helm uninstall <Releasename> -n <Namespace> --no-hooks