Recent testing failures

[squakez]

Recently we are having a couple of tests failing within the install workflow: https://github.com/apache/camel-k/actions/runs/3582747159/jobs/6027336970

❌ TestBasicSetup (1m13.94s)
No IntegrationPlatform resource in default namespace
...
Camel K Cluster Roles removed from cluster
Using alternative make directory on path ../../../../install
Using alternative make directory on path ../../../../install
setup_test.go:51:
Timed out after 60.003s.
Expected
<[]v1.Role | len:0, cap:0>: nil
To satisfy at least one of these matchers: [%!s(*matchers.HaveLenMatcher=&{7}) %!s(*matchers.HaveLenMatcher=&{8})]

❌ TestGlobalSetup (1m13.26s)
No IntegrationPlatform resource in default namespace
...
Using alternative make directory on path ../../../../install
Using alternative make directory on path ../../../../install
setup_test.go:78:
Timed out after 60.003s.
Expected
<[]v1.Role | len:7, cap:8>: [
{
TypeMeta: {Kind: "", APIVersion: ""},
ObjectMeta: {

[squakez]

I can see that even the successful tests of the same workflow report:

  Camel K Cluster Roles removed from cluster
  Using alternative make directory on path ../../../../install
  Using alternative make directory on path ../../../../install
  Using alternative make directory on path ../../../../install
      test_support.go:2239: Warning: some pods are still running in namespace "test-937a699e-8c7f-4ed4-8402-d82a9750213e" after deletion (1)

I wonder if this is the root cause.

[tadayosi]

This is a duplicate of #3772. I'm already working on it.

For now, what I see as the possible root cause is that kustomize taints kustomization.yaml under install/ each test run and it could be that the installation scripts are not idempotent, so the random order of test executions sometimes cause misconfigurations.