Prometheus graphs ignores metrics with empty container label

[RubenNL]

freelens doesn't calculate the pod cpu, memory and filesystem usage correctly. Here's an example of a single-container pod:

Before:

After:

I have this issue with those metrics:

container_cpu_usage_seconds_total
container_memory_working_set_bytes
container_fs_usage_bytes
container_fs_writes_bytes_total
container_fs_reads_bytes_total

Those are all metrics where the name starts with container_*, excluding container_network_* (Those never have a container attribute).

All other *-provider.injectable.ts have this same fix applied. I haven't been able to find why this one is different, so let's set them to the same queries.

Info about my cluster

I'm running k3s, with one master node and multiple "worker" nodes.
Prometheus is provided by the bitnami/kube-prometheus helm chart.

Tests

The tests have successfully ran on a testing PR on my own repo: RubenNL#1

[dex4er]

@RubenNL thanks a lot!

[robertobandini]

Thanks @RubenNL!