Stale image

[derrickburns]

It appears that the helm operator continues to look for updates to images after then have been removed from all existing helm releases.

I get:

ts=2019-02-23T01:49:53.028217228Z caller=warming.go:192 component=warmer canonical_name=index.docker.io/library/seagull auth={map[]} err="requesting tags: errors:\ndenied: requested access to the resource is denied\nunauthorized: authentication required\n"

However, all references to index.docker.io/library/seagull have been removed from all releases.

[stefanprodan]

The image scanning is being performed by Flux, the above are Flux logs not helm-op. Do you still see this error in Flux logs? The cache should be cleared after some time.

[derrickburns]

Thanks for the clarification regarding responsibility.

I will check to see if the cache was cleared. How long do you expect before the sweep?

[2opremio]

How long do you expect before the sweep?

It will take max(time_to_process_backlog), 1min), were time_to_process_backlog is the time it takes to go through all the repositories to sweep.

However, all references to index.docker.io/library/seagull have been removed from all releases.

Note that flux checks all the Kubernetes controllers existing in the cluster, not in your git repo.

@derrickburns my guess is that you have a controller in the cluster still using index.docker.io/library/seagull

@squaremo / @hiddeco why are we using the Cluster resources and not the Git repo manifest as the source of container image registries?

[stefanprodan]

@2opremio Flux has a readonly mode for Weave Cloud where we display all workloads running in your cluster. Scanning all the images is a feature that Weave Cloud needs.

[derrickburns]

Now that I understand flux better, I see that you are correct. There was probably an old helmrelease in the cluster. However, I had changed it in the repo. Derrick

…

On Feb 26, 2019, 1:54 AM -0800, Stefan Prodan ***@***.***>, wrote: @2opremio Flux has a readonly mode for Weave Cloud where we display all workloads running in your cluster. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.

[squaremo]

@derrickburns fluxd doesn't garbage collect resources at present (though see the ever-ripening #1442), so perhaps what's happened is that the resource name changed, and there's an out of date HelmRelease floating around.

Would you be able to check if there's a HelmRelease anywhere that you don't expect to be there? Mainly so we can post a plausible explanation, before tying up this issue.

[derrickburns]

There was an old HelmRelease. Thx!

[squaremo]

Thanks for confirming that @derrickburns. Closing as we have an explanation and a remedy.