GCP e2-custom-* instance type is not supported by our E2E test framework.
Now that testplatform have started using those instance types, we are seeing permafailing E2E job runs on our CPMS E2E periodic tests.

Error sample:

• [FAILED] [285.539 seconds]475ControlPlaneMachineSet Operator With an active ControlPlaneMachineSet and the instance type is changed [BeforeEach] should perform a rolling update [Periodic]476  [BeforeEach] /go/src/github.com/openshift/cluster-control-plane-machine-set-operator/test/e2e/periodic_test.go:39477  [It] /go/src/github.com/openshift/cluster-control-plane-machine-set-operator/test/e2e/periodic_test.go:43478479  [FAILED] provider spec should be updated with bigger instance size480  Expected success, but got an error:481      <*fmt.wrapError | 0xc000358380>: 482      failed to get next instance size: instance type did not match expected format: e2-custom-6-16384483      {484          msg: "failed to get next instance size: instance type did not match expected format: e2-custom-6-16384",485          err: <*fmt.wrapError | 0xc000358360>{486              msg: "instance type did not match expected format: e2-custom-6-16384",487              err: <*errors.errorString | 0xc0001489f0>{488                  s: "instance type did not match expected format",489              },490          },491      }

Always

1. Use e2-custom in GCP in a cluster, run CPMSO E2E periodics
2.
3.

Permafailing E2Es

Successful E2Es