Pods are not found

[alexanderniebuhr]

Describe the bug
We do have Pods which are created by another Pod (GitLab Runner Kubernetes Executer).
These Pods don't have a Deployment, but even while they are running, they are not found in krr.

How can these Pods also be checked.

Screenshots
If applicable, add screenshots to help explain your problem.

[obrienrobert]

I have a similar issue where pods are controlled by a service as opposed to a Deployment/StatefulSet/DaemonSet etc. It would be great if such pods could also be included in the result set.

[LeaveMyYard]

Hey! So, when starting this project we descided to skip recommendations for standalone pods
But if you guys ask, I think this is time to add this feature

[alexanderniebuhr]

We are also missing CronJob & Job

[LeaveMyYard]

@alexanderniebuhr Jobs should work as expected, but CronJobs we haven't added also
Will add those for future tasks

[mrclrchtr]

Thank you so much for this fantastic tool!

I was wondering today why the github runner wasn't showing up and then found this issue. This is probably related to that right?

[aantn]

@mrclrchtr can you try running krr with the prometheus-workload-loader branch and see if they show up there?

[mrclrchtr]

I tried the prometheus-workload-loader branch and unfortunately it doesn't show up.

[lictw]

The LimitRange is other case, so the deployments aren't have Resources values and leave them to shared/common values in LimitRange, but pods have requests and krr does not consider it. View with pods should help.

[hanleybrand]

I'm having a similar problem -- I ran krr on an airflow cluster, but only got back the pods that are in deployments, not the pods running the workflows which are standalone pods.

I can get currently running worker pods using
kubectl get pods -n airflow --selector 'component in (worker)'

NAME	READY	STATUS	RESTARTS	AGE
dap-init-or-sync-db-auff9ubk	1/1	Running	0	3m13s

but running
krr simple --cluster -n airflow --selector 'component in (worker)'
returns No objects available to scan.

[aantn]

@hanleybrand just to confirm, are they also missing on the prometheus-workload-loader branch?