VictoriaMetrics instead of Prometheus about coroot HOT 5 CLOSED

So in case someone is using victoria-metrics-k8s-stack helm chart to install VictoriaMetrics then its scraping configuration needs to be modified like this to make Coroot happy:

    kube-state-metrics:
      vmServiceScrape:
        spec:
          endpoints:
          - port: http
            honorLabels: true
            metricRelabelConfigs:
              - action: labeldrop
                regex: (container_id|image_id)

https://github.com/fenio/homelab/blob/main/cluster/apps/monitor/victoria-metrics/app/helmrelease.yaml#L69

from coroot.

We use coroot with victoriametrics without problems

from coroot.

@kirillyu thanks for reply
Did you have to somehow change VM configuration to make it working with Coroot?
I'm still digging into this and it seems that VM drops some labels that Coroot really would like to see.

E0324 16:20:44.957786 1 k8s.go:90] invalid 'kube_pod_info' metric: 'uid' label is empty

from coroot.

@fenio Yes, Coroot expects the uid label of thekube_pod_info metric. Additionally, please ensure that the VM has discovered the endpoints of coroot-node-agent and successfully scrapes them.

from coroot.

VM discovered these endpoints and it's scrapping them successfully.
But what's the easiest way to force VM to stop dropping that uid label from metrics?
I'm using default scrape_config configuration... do I have to make my own copy of it and remove all labeldrop instances referencing uid? Sounds crazy.

from coroot.