Gitlab k8s executor performance degradation on build+tests

Dear community,

could you please help me with an issue with GitLab (k8s executor) and my project?

Some context:

I have a project with tests, written in ruby. I have two pipelines: Jenkins and GitLab (in migration process).
On Jenkins VM, build+test duration is ~6 min, in GitLab the same job isn’t finished and killed by timeout.

Jenkins VM and GitLab k8s VM are identical by CPU/MEM/IO resources.

I’m wondering why it happens, so I isolated the k8s VM from other builds and run only my job, just to check if other containers impacts the build. But still no luck here - my build+test didn’t end with more than 1 hour.

I checked the limits in the pod - here is my configuration details for gitlab-runner.config.toml:

    cpu_request = "200m"
    memory_request = "300Mi"
    ephemeral_storage_request = "3Gi"

    cpu_limit = "4000m"
    memory_limit = "6000Mi"

And I can confirm that pod has the same limits in k8s dashboard.

I could see that build container inside the Pod doesn’t consume a CPU - only svc-0 does.

I assume that’s why my tests running so long…

Could anybody help me with the suggestion (where to dig :smiley: ) - really appreciate anything :slight_smile:

Best regards,

Please disregard it - it was internal issue with tests.

1 Like