Fix integration testing of harbor-trivy StatefulSet on GitHub runenrs.

[aznashwan]

Summary

Currently the integration tests checks all components defined in the upstream Harbor Helm chart except for the harbor-trivy StatefulSet, which consistently seems to have issues getting its PVC allocated for it on the GitHub runners with errors such as:

# https://github.com/aznashwan/harbor-rocks/actions/runs/10151954257/job/28078474328#step:8:2728
$ kubectl describe pod harbor-trivy-0
...
Events:
  Type     Reason       Age                   From               Message
  ----     ------       ----                  ----               -------
  Normal   Scheduled    5m29s                 default-scheduler  Successfully assigned default/harbor-trivy-0 to k8s-integration-89872c-3
  Warning  FailedMount  79s (x10 over 5m29s)  kubelet            MountVolume.MountDevice failed for volume "pvc-aa087c05-b90b-4bcb-b4be-a7399d2c01c6" : rpc error: code = Unknown desc = Exception calling application: Command 'losetup --direct-io=on -f /data/pvc-aa087c05-b90b-4bcb-b4be-a7399d2c01c6/disk.img' returned non-zero exit status 1.

Or:

# https://github.com/aznashwan/harbor-rocks/actions/runs/10148154466/job/28068574283#step:8:7532
$ kubectl describe pod harbor-trivy-0
...
Events:
  Type     Reason            Age                    From               Message
  ----     ------            ----                   ----               -------
  Warning  FailedScheduling  5m43s                  default-scheduler  running PreBind plugin "VolumeBinding": binding volumes: provisioning failed for PVC "data-harbor-trivy-0"
  Warning  FailedScheduling  5m36s (x2 over 5m40s)  default-scheduler  running PreBind plugin "VolumeBinding": binding volumes: provisioning failed for PVC "data-harbor-trivy-0"

What Should Happen Instead?

The integration tests should run fine on the GitHub amd64 runners, as they seem to consistently run fine in local setups when straight-up copy-pasting the exact same helm install command.

Reproduction Steps

Enable the checks for the harbor-trivy StatefulSet in the integration tests and re-run the pull_request.yaml workflow on this repo on the default Ubuntu-based GitHub AMD64 runners.

Can you suggest a fix?

Might be related to the number of loopback devices defined in the LXD profile used for integration testing.

Are you interested in contributing with a fix?

In the future, yes, but I currently need to focus on other things.