Hide Forgot
Bug 1920209 made multus rollouts slightly faster, but they're still serial. We want maxUnavailable: 10%, so we scale better on clusters with large node-counts. Checking a recent 4.8 CI release: $ curl -s https://gcsweb-ci.apps.ci.l2s4.p1.openshiftapps.com/gcs/origin-ci-test/logs/release-openshift-origin-installer-e2e-gcp-4.8/1364411741282242560/artifacts/e2e-gcp/daemonsets.json | jq -r '.items[] | select(.spec.template.spec.nodeSelector["node-role.kubernetes.io/master"] != "" and .spec.updateStrategy.rollingUpdate.maxUnavailable != "10%") | .metadata.namespace + " " + .metadata.name + " " + (.spec.template.spec.nodeSelector | tostring) + " " + (.spec.updateStrategy | tostring)' ... openshift-multus multus {"kubernetes.io/os":"linux"} {"rollingUpdate":{"maxUnavailable":1},"type":"RollingUpdate"} openshift-multus network-metrics-daemon {"kubernetes.io/os":"linux"} {"rollingUpdate":{"maxUnavailable":1},"type":"RollingUpdate"} ... And we want to empty that query out before 4.8 GAs.
cluster-network-operator#992 looks like it will address openshift-multus/multus, but per comment 0, this bug is also asking for openshift-multus/network-metrics-daemon to be bumped (around [1]). [1]: https://github.com/openshift/cluster-network-operator/blob/dfea1dc53ed82f5e4a4e85b7e8e863af3e4bd54a/bindata/network/network-metrics/001-daemonset.yaml#L17
Thanks W. Trevor, also for the file pointer. I've gone ahead and updated cluster-network-operator#992 to update the network-metrics-daemon as well.
Used latest from https://gcsweb-ci.apps.ci.l2s4.p1.openshiftapps.com/gcs/origin-ci-test/logs/release-openshift-origin-installer-e2e-gcp-4.8, the curl result is still not empty. [weliang@weliang ~]$ curl -s https://gcsweb-ci.apps.ci.l2s4.p1.openshiftapps.com/gcs/origin-ci-test/logs/release-openshift-origin-installer-e2e-gcp-4.8/1366818372670459904/artifacts/e2e-gcp/daemonsets.json | jq -r '.items[] | select(.spec.template.spec.nodeSelector["node-role.kubernetes.io/master"] != "" and .spec.updateStrategy.rollingUpdate.maxUnavailable != "10%") | .metadata.namespace + " " + .metadata.name + " " + (.spec.template.spec.nodeSelector | tostring) + " " + (.spec.updateStrategy | tostring)' ... openshift-cluster-csi-drivers gcp-pd-csi-driver-node {"kubernetes.io/os":"linux"} {"rollingUpdate":{"maxUnavailable":1},"type":"RollingUpdate"} openshift-image-registry node-ca {"kubernetes.io/os":"linux"} {"rollingUpdate":{"maxUnavailable":1},"type":"RollingUpdate"} openshift-multus multus {"kubernetes.io/os":"linux"} {"rollingUpdate":{"maxUnavailable":1},"type":"RollingUpdate"} openshift-multus network-metrics-daemon {"kubernetes.io/os":"linux"} {"rollingUpdate":{"maxUnavailable":1},"type":"RollingUpdate"} openshift-network-diagnostics network-check-target {"beta.kubernetes.io/os":"linux"} {"rollingUpdate":{"maxUnavailable":1},"type":"RollingUpdate"} openshift-sdn ovs {"kubernetes.io/os":"linux"} {"rollingUpdate":{"maxUnavailable":1},"type":"RollingUpdate"} openshift-sdn sdn {"kubernetes.io/os":"linux"} {"rollingUpdate":{"maxUnavailable":1},"type":"RollingUpdate"} ... @dosmith Is above curl result expected?
The origin tests allow 33% as well as 10%, but your curl is still showing 1, which we don't want. Looking at 4.8.0-0.nightly-2021-03-18-075013 [1,2], I see: $ curl -s https://gcsweb-ci.apps.ci.l2s4.p1.openshiftapps.com/gcs/origin-ci-test/logs/release-openshift-ocp-installer-e2e-aws-4.8/1372456130126548992/artifacts/e2e-aws/daemonsets.json | jq -r '.items[] | select(.spec.template.spec.nodeSelector["node-role.kubernetes.io/master"] != "" and .spec.updateStrategy.rollingUpdate.maxUnavailable != "10%") | .metadata.namespace + " " + .metadata.name + " " + ( .spec.template.spec.nodeSelector | tostring) + " " + (.spec.updateStrategy | tostring)' openshift-multus network-metrics-daemon {"kubernetes.io/os":"linux"} {"rollingUpdate":{"maxUnavailable":"33%"},"type":"RollingUpdate"} So that's not empty, but the 33% is fine. [1]: https://amd64.ocp.releases.ci.openshift.org/releasestream/4.8.0-0.nightly/release/4.8.0-0.nightly-2021-03-18-075013 [2]: https://prow.ci.openshift.org/view/gs/origin-ci-test/logs/release-openshift-ocp-installer-e2e-aws-4.8/1372456130126548992
I'm seeing that the Multus daemonset is 10% (which is I think the conclusion that we came to in the PR), I'm using a cluster-bot ci cluster (4.8.0-0.ci-2021-03-19-084923), the steps I took to check are: ``` oc project openshift-multus oc get daemonset multus -o yaml | grep -A3 -Pi "[^:]updateStrategy:" updateStrategy: rollingUpdate: maxUnavailable: 10% type: RollingUpdate ``` ...Oh, looks like Weibin since verified *thumbs up* At any rate, just an fyi then.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (Moderate: OpenShift Container Platform 4.8.2 bug fix and security update), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHSA-2021:2438