The job periodic-ci-openshift-multiarch-master-nightly-4.15-ocp-e2e-ovn-remote-libvirt-s390x-heterogeneous currently does not skip tests that are not yet enabled for multi-arch compute setups and are skipped on other platforms, for examle on Power: https://github.com/openshift/release/blob/master/ci-operator/config/openshift/multiarch/openshift-multiarch-master__nightly-4.15.yaml#L168-L181

These failures can be seen e.g. here:

• https://prow.ci.openshift.org/view/gs/test-platform-results/pr-logs/pull/openshift_release/48112/rehearse-48112-periodic-ci-openshift-multiarch-master-nightly-4.15-ocp-e2e-ovn-remote-libvirt-s390x-heterogeneous/1762490410581626880
• https://prow.ci.openshift.org/view/gs/test-platform-results/pr-logs/pull/openshift_release/48112/rehearse-48112-periodic-ci-openshift-multiarch-master-nightly-4.15-ocp-e2e-ovn-remote-libvirt-s390x-heterogeneous/1760342455980396544
• https://prow.ci.openshift.org/view/gs/test-platform-results/pr-logs/pull/openshift_release/48112/rehearse-48112-periodic-ci-openshift-multiarch-master-nightly-4.15-ocp-e2e-ovn-remote-libvirt-s390x-heterogeneous/

These skips are implemented in the ref openshift-e2e-test (https://github.com/openshift/release/blob/master/ci-operator/step-registry/openshift/e2e/test/openshift-e2e-test-commands.sh#L288-L294) but not in openshift-e2e-libvirt-test (https://github.com/openshift/release/blob/master/ci-operator/step-registry/openshift/e2e/libvirt/test/openshift-e2e-libvirt-test-commands.sh)

 
Possible fixes are improving the connection or introducing some leeway for pathological events of this type in this environments.

Suggested test skips:

TEST_SKIPS: deploymentconfigs| should expose cluster services outside the cluster|
  FIPS TestFIPS| Multi-stage image builds should succeed| Optimized image
  builds should succeed| build can reference a cluster service| custom build
  with buildah| oc new-app should succeed| prune builds based on settings|
  s2i build with a root| verify /run filesystem contents| oc can run| oc
  debug| oc idle| Pods cannot access| Image append should create| Image
  extract should extract| Image info should display| Image layer subresource|
  oc tag should change image| when installed on the cluster should| OpenShift
  alerting rules| The HAProxy router should| egressrouter cni resources|
  pod should start| pod sysctls| Managed cluster should have same number of
  Machines and Nodes | Managed cluster should have machine resources | should
  not have pod creation failures during install | DNS should answer queries
  using the local DNS endpoint | not repeat pathologically for ns/openshift-etcd
  | egressFirewall should have no impact outside its namespace

Failing tests:

• [sig-builds][Feature:Builds] oc new-app should succeed with a --name of 58 characters [apigroup:build.openshift.io] [Skipped:Disconnected] [Skipped:Proxy] [Suite:openshift/conformance/parallel]
• [sig-apps][Feature:DeploymentConfig] deploymentconfigs won't deploy RC with unresolved images when patched with empty image [apigroup:apps.openshift.io] [Skipped:Disconnected] [Suite:openshift/conformance/parallel] 
• [sig-apps][Feature:DeploymentConfig] deploymentconfigs when changing image change trigger should successfully trigger from an updated image [apigroup:apps.openshift.io][apigroup:image.openshift.io] [Skipped:Disconnected] [Suite:openshift/conformance/parallel]
• [sig-builds][Feature:Builds] s2i build with a root user image should create a root build and pass with a privileged SCC [apigroup:build.openshift.io] [Skipped:Disconnected] [Suite:openshift/conformance/parallel]

Not covered by these skips:

• [sig-network][Feature:EgressFirewall] when using openshift ovn-kubernetes should ensure egressfirewall is created [Suite:openshift/conformance/parallel] - Fixed by https://issues.redhat.com/browse/OCPBUGS-20487 for 4.16, should be skipped for 4.15
• [sig-builds][Feature:Builds] s2i build with a quota Building from a template should create an s2i build with a quota and run it [apigroup:build.openshift.io] [Skipped:Disconnected] [Suite:openshift/conformance/parallel] - Single failure in early run, have to check if it occurs again or was one off.