Description of problem: must-gather pull-policy is "IfNotPresent". This sometimes ends in using a non-latest build of must-gather image, e.g.: When running: $ oc adm must-gather --image=registry-proxy.engineering.redhat.com/rh-osbs/container-native-virtualization-cnv-must-gather-rhel8:v2.4.0 --dest-dir=/home/cnv-qe-jenkins/yossi/mg Let's assume there's already an image of v2.4.0 in the cluster (which remained there from a previous run of must-gather), e.g. from build #42, hence v2.4.0-42. If the build *in the image repo* advanced, and now the latest there is #43 (v2.4.0-43), then the "IfNotPresent" pull-policy would prevent pulling the latest build. Version-Release number of selected component (if applicable): $ oc version Client Version: 4.5.0-rc.1 Server Version: 4.5.0-rc.5 Kubernetes Version: v1.18.3+f291db1 How reproducible: Always, although it is not easy to catch, as a user. There should be a significant difference between the existing, outdated image in the cluster and the one in the repo. Steps to Reproduce: 1. Run a simple must-gather action, e.g.: $ oc adm must-gather --image=registry-proxy.engineering.redhat.com/rh-osbs/container-native-virtualization-cnv-must-gather-rhel8:v2.4.0 --dest-dir=/home/cnv-qe-jenkins/yossi/mg Actual results: If an older image if the same version (v2.4.0 in this example) exists in the cluster - no new image will be pulled from the repo, even if the image in the repo is a newer build of the same version. Expected results: Always use the latest version (and build) of an image. Additional info: This was found whne using a CNV (Openshift Virtualization) must-gather image, but was debugged and found to be an issue the stems from the OCP must-gather pull-policy (in bz#1859972).
The linked PR should merge this sprint to resolve this.
Verified bug in the payload below and i see that the imagepullpolicy has been changed to Always. [ramakasturinarra@dhcp35-60 verification-tests]$ oc get clusterversion NAME VERSION AVAILABLE PROGRESSING SINCE STATUS version 4.6.0-0.nightly-2020-08-26-032807 True False 4h6m Cluster version is 4.6.0-0.nightly-2020-08-26-032807 [ramakasturinarra@dhcp35-60 verification-tests]$ oc version -o yaml clientVersion: buildDate: "2020-08-26T05:14:57Z" compiler: gc gitCommit: b08e3b04f1b19d9a7265b4c44372b44090262a87 gitTreeState: clean gitVersion: 4.6.0-202008260439.p0-b08e3b0 goVersion: go1.14.4 major: "" minor: "" platform: linux/amd64 openshiftVersion: 4.6.0-0.nightly-2020-08-26-032807 serverVersion: buildDate: "2020-08-25T13:43:16Z" compiler: gc gitCommit: aaf4ce1cf2d4201a23e332030ccfeab6c17351fd gitTreeState: dirty gitVersion: v1.19.0-rc.2+aaf4ce1-dirty goVersion: go1.14.4 major: "1" minor: 19+ platform: linux/amd64 [ramakasturinarra@dhcp35-60 verification-tests]$ oc get pod must-gather-sm42d -o yaml | grep imagePullPolicy f:imagePullPolicy: {} f:imagePullPolicy: {} imagePullPolicy: Always imagePullPolicy: Always Based on the above moving the bug to verified state
Looks like the fix for this was reverted - https://github.com/openshift/oc/commit/de7c254f96746682bd7bbf7006a1f4da1751c114
Good catch! Thanks, Yuval. I submitted a needInfo question in the bug that is the source of this fix (https://bugzilla.redhat.com/show_bug.cgi?id=1874856#c5), with an ask to revert this. Let's wait and see how that turns out.
Well, you can look at the discussion on the PR https://github.com/openshift/oc/pull/549 I guess this one will be resolved as WONTFIX :/
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 (OpenShift Container Platform 4.6 GA Images), 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/RHBA-2020:4196