[Bug] Failed to pull image "temporalio/admin-tools:1.24.0"

[CarlosBorroto]

What are you really trying to do?

I'm upgrading to temporal-0.39.0 and admintools pods are failing to start because of what looks like a mismatch in docker tags used by the chart and what is present in docker.io.

Describe the bug

Looking at docker.io, it looks like Temporal is now using tags that look like v1.24.0-0 instead of 1.24.0. This is causing all pods to try to pull the incorrect image.

│ Events:                                                                                                                                                                                                    │
│   Type     Reason     Age                    From               Message                                                                                                                                    │
│   ----     ------     ----                   ----               -------                                                                                                                                    │
│   Normal   Scheduled  5m38s                  default-scheduler  Successfully assigned temporal/temporal-admintools-647697cfbf-q9ngh to ip-10-164-17-207.ec2.internal                                       │
│   Normal   Pulling    4m4s (x4 over 5m38s)   kubelet            Pulling image "temporalio/admin-tools:1.24.0"                                                                                              │
│   Warning  Failed     4m4s (x4 over 5m38s)   kubelet            Failed to pull image "temporalio/admin-tools:1.24.0": rpc error: code = NotFound desc = failed to pull and unpack image "docker.io/tempora │
│ lio/admin-tools:1.24.0": failed to resolve reference "docker.io/temporalio/admin-tools:1.24.0": docker.io/temporalio/admin-tools:1.24.0: not found                                                         │
│   Warning  Failed     4m4s (x4 over 5m38s)   kubelet            Error: ErrImagePull                                                                                                                        │
│   Warning  Failed     3m49s (x6 over 5m38s)  kubelet            Error: ImagePullBackOff                                                                                                                    │
│   Normal   BackOff    31s (x20 over 5m38s)   kubelet            Back-off pulling image "temporalio/admin-tools:1.24.0"

Minimal Reproduction

Deploy temporal-0.39.0 helm chart version.

[CarlosBorroto]

Also reported here. It might be the actual bug is in the ci in that repo that should push this tag.

[alexshtin]

I would suggest you to always specify image tags explicitly when using helm charts. The right admin-tools tag for 1.24.0 is 1.24.0-tctl-1.18.1-cli-0.12.0.

We gonna remove defaults for image tags in helm charts.

[CarlosBorroto]

@alexshtin are you recommending even .Values.server.image.tag which defaults to .Chart.AppVersion should be always specified? Or, just the extra container images like .Values.admintools.image.tag and .Values.web.image.tag? I can see why having admintools defaulting to .Chart.AppVersion is a bad idea seeing they don't share the same image. However, for the server deployment, I don't see the need to explicitly specify unless you are planning to stop updating .Chart.AppVersion.

We will start specifying all 3 tags moving forward, although, I would honestly prefer if the helm charts owners would take on providing good defaults for those. We really have no idea which admintools or web image will be supported by a particular chart version.

[ascknx]

With helm chart 0.42.0 the same happened with the temporalio/ui image. This is a really bad user experience. Can you please specify the working versions in the helm chart?