Bug 2026806

I closed this CURRENTRELEASE, but then realized that we should see improvements in the 4.7 -> ... -> 4.10 job [1] now that the master/4.10 PR has landed with this bug.  I'm agnostic about whether we stay in CURRENTRELEASE or move back to MODIFIED/ON_QA as we wait for that to get some new runs.  I've also opened the 4.9 backport bug 2027929, and I'm agnostic about how long we cook in 4.10 before moving ahead and landing that backport.

[1]: https://testgrid.k8s.io/redhat-openshift-ocp-release-4.10-informing#release-openshift-origin-installer-e2e-aws-upgrade-4.7-to-4.8-to-4.9-to-4.10-ci

*** Bug 2028761 has been marked as a duplicate of this bug. ***

this e2e is failing on direct install0f 4.8.23 as well on upgrade job from 4.8.22 to 4.8.23 for ppc64le 
" [sig-cluster-lifecycle] TestAdminAck should succeed [Suite:openshift/conformance/parallel]"

S390x Upgrade jobs for 4.9 from 4.8 is continuously failing with test "disruption_tests: [bz-Cluster Version Operator] Verify presence of admin ack gate blocks upgrade until acknowledged" 

https://prow.ci.openshift.org/view/gs/origin-ci-test/logs/periodic-ci-openshift-multiarch-master-nightly-4.9-upgrade-from-nightly-4.8-ocp-remote-libvirt-s390x/1467690345897660416

Checking a 4.7 -> ... -> 4.10 run, now that the third PR is in [1,2].  It has Jack's map-nil fix:

  $ curl -s https://gcsweb-ci.apps.ci.l2s4.p1.openshiftapps.com/gcs/origin-ci-test/logs/release-openshift-origin-installer-e2e-aws-upgrade-4.7-to-4.8-to-4.9-to-4.10-ci/1467585900341891072/artifacts/release/artifacts/release-images-latest | jq -r '.spec.tags[] | select(.name == "tests").annotations["io.openshift.build.commit.id"]'
  bc643fa990bef62359eeaf8c54e1aa475f642193
  $ git --no-pager log --oneline -1 bc643fa990bef62359eeaf8c54e1aa475f642193
  bc643fa990 Merge pull request #26668 from jottofar/check-for-nil-map

But there's still a failure due to an DNS/networking hiccup:

  $ curl -s https://storage.googleapis.com/origin-ci-test/logs/release-openshift-origin-installer-e2e-aws-upgrade-4.7-to-4.8-to-4.9-to-4.10-ci/1467585900341891072/build-log.txt | grep FAIL:
  Dec  6 00:00:10.440: FAIL: Error accessing configmap openshift-config-managed/admin-gates: Get "https://api.ci-op-g2m38jp7-eafe9.origin-ci-int-aws.dev.rhcloud.com:6443/api/v1/namespaces/openshift-config-managed/configmaps/admin-gates": dial tcp: lookup api.ci-op-g2m38jp7-eafe9.origin-ci-int-aws.dev.rhcloud.com on 172.30.0.10:53: no such host

We may want to relax the test-case a bit so that it comes back and tries again when that sort of thing happens, instead of calling framework.Fail.

[1]: https://testgrid.k8s.io/redhat-openshift-ocp-release-4.10-informing#release-openshift-origin-installer-e2e-aws-upgrade-4.7-to-4.8-to-4.9-to-4.10-ci
[2]: https://prow.ci.openshift.org/view/gs/origin-ci-test/logs/release-openshift-origin-installer-e2e-aws-upgrade-4.7-to-4.8-to-4.9-to-4.10-ci/1467585900341891072

The failures mentioned in comment 6 are from adminack.go line 193, and that's the nil-map thing Jack's fixed in master in origin#26668:

$ git cat-file -p origin/release-4.9:test/extended/util/adminack.go | grep -n . | grep -5 ^193:
188:func setAdminGate(ctx context.Context, gateName string, gateValue string, oc *CLI) string {
189:    ackCm, errMsg := getAdminAcksConfigMap(ctx, oc)
190:    if len(errMsg) != 0 {
191:            framework.Failf(errMsg)
192:    }
193:    ackCm.Data[gateName] = gateValue
194:    _, err := oc.AdminKubeClient().CoreV1().ConfigMaps("openshift-config").Update(ctx, ackCm, metav1.UpdateOptions{})
195:    if err != nil {
196:            return fmt.Sprintf("Unable to update configmap openshift-config/admin-acks, err=%v.", err)
197:    }
198:    return ""

I'm punting the "no such host" hiccup from comment 8 to follow up work, and marking this CLOSED CURRENTRELEASE based on the other changes it made fixing issues we had been seeing in CI.