Clear helm cache from rancher


I’m exploring Helm charts for usage. I created a test app catalog, and upload a chart. But there is a bugs in chart, and it’s failed to install. I fixed bugs, upload new chart with same name and version, refresh app catalog in rancher UI. After it, i try to deploy this app again, and it’s deploy from the old chart. I think, there are some cache in rancher… How can i force clear it?

Rancher version 2.2.7.

1 Like

Although I believe this issue is already fixed by now, we had that “problem” with earlier versions of Rancher 2 and our own catalog too.
A workaround for us was to simply connect to the rancher2 cluster and update the catalog-cache on each rancher-server pod via git.
Here’s how we did it:
for name in $(kc get pods | grep rancher-|cut -d ’ ’ -f 1); do kc exec -it $name – sh -c “cd management-state/catalog-cache/ENTER_CACHE_ID_HERE && git pull”; done

Please note that this is just a workaround and I think that there’s already a solution for the issue implemented, since we are running version 2.2.5 and do no longer need to update the catalog manually.

Same issue on v2.3.3 . I need to push an entirely new version to see changes. If I push to the same version, but with new content, then Rancher still uses the old Chart of the same version.

Rancher caches helm charts based on their version. If the chart version doesn’t change, rancher assumes the chart hasn’t changed. There is no guarantee that a forceRefresh will update your chart in rancher if it has the same chart version. I describe a workaround to this issue in the provided post. Hope it helps!

I’m using Rancher 2.2.5 and I have the same issue. I’m trying to just change app-readme. The readme is updated when I increase chart version but won’t update when I push to an existing version. Refresh catalog is not helping either.