Helm chart installation times out and starts to consume huge amount of CPU

[margorczynski]

Bug description

After trying to install Superset on my local minikube k8s cluster I get the following:

• chart installation times out or hangs up
• a gunicorn process is spawned that along with the dockerd consumes huge amounts of CPU (16-core Ryzen 7950X)

How to reproduce the bug

1. helm upgrade --install superset superset/superset --namespace=mynamespace

Screenshots/recordings

Superset version

3.1.2

Python version

I don't know

Node version

I don't know

Browser

Not applicable

Additional context

No other Helm installation on the cluster causes such issues

Checklist

• I have searched Superset docs and Slack and didn't find a solution to my problem.
• I have searched the GitHub issue tracker and didn't find a similar bug report.
• I have checked Superset's logs for errors and if I found a relevant Python stacktrace, I included it here as text in the "additional context" section.

[villebro]

@margorczynski please add which version of the Helm chart you're running.

[margorczynski]

It should be the latest considering the command: @villebro

[margorczynski]

From what I see in Kubernetes the superset-init-db job constantly spawns failing pods that fail with the following:

I didn't see any config value to set the secret in the Helm chart. And if it would be required shouldn't the install fail because of a missing parameter?

[margorczynski]

After running the install with values that override SECRET_KEY it install correctly. So it looks like I misunderstood the guide. But still it says: Default security settings and passwords are included but you SHOULD override those with your own, in particular: but it seems SHOULD should be replace by MUST for it to work and mention it before the install command.

Maybe an explicit mention before the helm install command? Or is it possible to generate it automatically and put it into Kubernetes secrets?

Still, after it install and 10 minutes pass I still see a sharp increase in resource usage and Kubernetes becomes unresponsive:

I will try to play around with the settings of minikube and Docker as there might be some issue there on my side but other charts like MinIO or Grafana didn't cause such problems

EDIT: After re-creating minikube with more resources (8 CPU, 16GB memory) it starts and works normally, can be closed with maybe some update to the Helm install docs

[rusackas]

it seems SHOULD should be replace by MUST for it to work

I'll open a PR for this... but please feel free to make additional docs PRs or let me know what changes you think need to be made. I'll see the PR to close this issue when merged.