CASMTRIAGE-7221: Sat unable to pull image #3673
Conversation
|
Testing output from starlord
|
annapoorna-s-alt
requested review from
haasken-hpe,
shivaprasad-metimath and
vinanti-p
There was a problem hiding this comment.
I think the zypper update needs to happen later in this script.
Can you please clarify how this was tested? Was the setup-nexus.sh script just re-run after the problem was observed?
If that's the case, then I assume the lib/setup-nexus.sh script had already been run (by prerequisites.sh). So the updated cray-sat-podman RPM would have already been uploaded to Nexus by the previous invocation of this script. That would explain why the added zypper update command updated the cray-sat-podman RPM to the latest version despite being before the nexus-upload commands that upload RPMs to the package repos in Nexus.
lib/setup-nexus.sh
Outdated
| @@ -31,6 +31,9 @@ skopeo-copy "${sat_image}:${sat_version}" "${sat_image}:csm-latest" | |||
| mkdir -p /opt/cray/etc/sat | |||
| echo "${sat_version}" > /opt/cray/etc/sat/version | |||
|
|
|||
| # Update the cray-sat-podman package to ensureversion consistency | |||
| zypper update cray-sat-podman | |||
There was a problem hiding this comment.
Upon closer review, I think this is the wrong location for this command. It looks like lines 40-46 below are the lines that actually upload the new RPMs from this CSM release into Nexus. We want to run the zypper update after that happens. I'm not exactly sure which repo has the newer cray-sat-podman RPM in it, so we could just run it after all the nexus-upload commands.
There was a problem hiding this comment.
I took a look at Nexus on starlord, and it looks like it's included in the csm-${RELEASE_VERSION}-embedded repository. I think that's where packages which are also already a part of the new CSM management node images are located, if I recall correctly.
So the zypper update would need to be done after line 46.
There was a problem hiding this comment.
Can you please clarify how this was tested? Was the setup-nexus.sh script just re-run after the problem was observed?
yes I observed the issue on starlord after reporting from QE while upgrading the system from 1.5 to 1.6. With the changes I reran the setup-nexus.sh script. Thats the output I added. As per your comment I addressed to have the changes after uploading the new RPMs from CSM release into Nexus.
Sat unable to pll the image on Tyr after a CSM 1.6.0-alpha.55 upgrade Solution: Add a zypper update cray-sat-podman after CSM 1.6 RPMs are uploaded to Nexus
annapoorna-s-alt
force-pushed
the
CASMTRIAGE-7221
branch
from
190efc6
to
026fc22
Compare
|
@rustydb, @mitcharf, @mtupitsyn, does it seem reasonable to you to add a |
|
This is obsoleted by https://github.com/Cray-HPE/docs-csm/pull/5486/files That PR adds a call to |
Summary and Scope
Sat unable to pull the image on Tyr after a CSM 1.6.0-alpha.55 upgrade
Solution: Add a zypper update cray-sat-podman after CSM 1.6 RPMs are uploaded to Nexus
Issues and Related PRs
Resolves CASMTRIAGE-7221.
Testing
List the environments in which these changes were tested.
Tested on:
starlord
Test description:
Test the setup-nexus.sh script of management nodes rollout stage when upgrading starlord from sm 1.5 to to 1.6
Risks and Mitigations
Low
Pull Request Checklist