Has anyone successfully completed the lab for section 6: Prometheus?

djarid

The pre-reqs section is a mess... there are references to commands that have not been run, I've muddled through to the point that I'm trying to follow this instruction:
helm upgrade cilium cilium/cilium --namespace=kube-system --reuse-values --set prometheus.enabled=true --set operator.prometheus.enabled=true
and get the error message

Error: UPGRADE FAILED: template: cilium/templates/hubble/servicemonitor.yaml:1:76: executing "cilium/templates/hubble/servicemonitor.yaml" at <.Values.hubble.metrics.dynamic.enabled>: nil pointer evaluating interface {}.enabled

Now fixing this setting leads me down a rabbit hole of related and unrelated settings that need to be applied (I stopped at the 10th error for nil pointer values)

I have version pinned cilium to 1.16.3 as referenced in the docs

fazlur.khan

Dear @djarid ,

From the error it looks like Helm tried to render the hubble/servicemonitor.yaml template but failed because it expected .Values.hubble.metrics.dynamic.enabled to be defined, and it found nil instead. In simpler terms, the Helm chart expects this value in your values.yaml or overrides, but it’s missing.

Try setting the value to false - helm upgrade --install cilium cilium/cilium --set hubble.metrics.dynamic.enabled=false

Meanwhile I will inform the author/maintainers of the course to validate and make changes to the lab if necessary. Thank you for your patience...

djarid

Thanks, the problem is there are around two dozen nil entries where objects are expected (with descendent values) so the whole thing took me several hours to configure. I ended up using the helm values command to create the correct values.yaml and editing that.

There are many assumptions like this throughout the course. It really does need a thorough review (particluarly the assumption in one section of v1.13 and v1.16 in others

djarid

I'd have completed this in 2 days instead of the 4.5 it took to work around all the errata