images: add python3-pcp for RHEL images

[jelly]

The Python rewrite of cockpit-pcp is underway and to help development and testing add the new runtime dependency.

• image-refresh rhel-9-5
• fix: adapt to the new state of insights-client.service on failure candlepin/subscription-manager-cockpit#80

[cockpituous]

image-refresh rhel-10-0 done: https://github.com/cockpit-project/bots/commits/image-refresh-rhel-10-0-20240822-150951

[cockpituous]

Success. Log: https://cockpit-logs.us-east-1.linodeobjects.com/image-refresh-rhel-10-0-85478484-20240822-145427/log.html

[cockpituous]

image-refresh rhel-10-0 done: https://github.com/cockpit-project/bots/commits/image-refresh-rhel-10-0-20240822-151111

[cockpituous]

Success. Log: https://cockpit-logs.us-east-1.linodeobjects.com/image-refresh-rhel-10-0-85478484-20240822-145430/log.html

[cockpituous]

image-refresh rhel-9-5 done: https://github.com/cockpit-project/bots/commits/image-refresh-rhel-9-5-20240822-152713

[cockpituous]

Success. Log: https://cockpit-logs.us-east-1.linodeobjects.com/image-refresh-rhel-9-5-85478484-20240822-150957/log.html

[cockpituous]

image-refresh centos-10 done: https://github.com/cockpit-project/bots/commits/image-refresh-centos-10-20240822-154802

[cockpituous]

Success. Log: https://cockpit-logs.us-east-1.linodeobjects.com/image-refresh-centos-10-85478484-20240822-153056/log.html

[martinpitt]

Just to make sure, you only want this as a runtime dependency, not a build-time one (BuildRequires:), right? I.e. pytest should be able to skip the pcp unit tests. (This is fine for me)

Tests are an absolute disaster.. 😢 this is definitively a "yell very loudly on Monday" thing.

[jelly]

Just to make sure, you only want this as a runtime dependency, not a build-time one (BuildRequires:), right? I.e. pytest should be able to skip the pcp unit tests. (This is fine for me)

Yes

Tests are an absolute disaster.. 😢 this is definitively a "yell very loudly on Monday" thing.

Ugh, kinda feared that. Should I split it up?

[martinpitt]

@ptoscano Can you please have a look at the three sub-man regressions on RHEL 10? sub-man, sub-man-cockpit, and on rhel-9-5 ? Same test, and same failed assertion, but slightly different failure mode -- one screenshot has a popup alert ("Failure in name resolution"), the other doesn't.

@jelly I take a look at the c10s failures first, they also break TF runs. As this is such a mess, I suggest to only take one image refresh here, and handle the other images in separate PRs.

[martinpitt]

Most of the centos-10 regressions are explained by the new kernel breaking journal logging. I reported this as https://issues.redhat.com/browse/RHEL-56025 . This has a reasonably straightforward workaround, however, it's not the full story as the TF failures show. However, let's do another refresh with that fix, and see where we land.

[cockpituous]

image-refresh centos-10 done: https://github.com/cockpit-project/bots/commits/image-refresh-centos-10-20240826-032738

[cockpituous]

Success. Log: https://cockpit-logs.us-east-1.linodeobjects.com/image-refresh-centos-10-4bce68a3-20240826-030943/log.html

[cockpituous]

image-refresh rhel-10-0 done: https://github.com/cockpit-project/bots/commits/image-refresh-rhel-10-0-20240826-034313

[cockpituous]

image-refresh centos-10 done: https://github.com/cockpit-project/bots/commits/image-refresh-centos-10-20240826-075906

[cockpituous]

Success. Log: https://cockpit-logs.us-east-1.linodeobjects.com/image-refresh-centos-10-4779053a-20240826-074228/log.html

[cockpituous]

image-refresh rhel-10-0 done: https://github.com/cockpit-project/bots/commits/image-refresh-rhel-10-0-20240826-081522

[cockpituous]

Success. Log: https://cockpit-logs.us-east-1.linodeobjects.com/image-refresh-rhel-10-0-4779053a-20240826-075913/log.html

[martinpitt]

OK, there's no sense in trying to hack around this. This is broken beyond our reach. https://issues.redhat.com/browse/RHEL-54549 is likely a duplicate and points to some more root causes. So this is just broken for now. I'll still keep the cleanup commit, and let's consider RHEL 10 as "broken" and rebuilds as blocked. Let's instead try to refresh rhel 9.5 so that @jelly has something to play with for pcp?

[cockpituous]

image-refresh rhel-9-5 done: https://github.com/cockpit-project/bots/commits/image-refresh-rhel-9-5-20240826-084927

[cockpituous]

Success. Log: https://cockpit-logs.us-east-1.linodeobjects.com/image-refresh-rhel-9-5-ca684ffd-20240826-083232/log.html

[martinpitt]

So it's just that sub-man regression now. @ptoscano can you please have a look?

[ptoscano]

So it's just that sub-man regression now. @ptoscano can you please have a look?

It's due to recent changes in insights-client that appeared few days ago in RHEL composes; see candlepin/subscription-manager-cockpit#80 for a longer explanation and hopefully a fix (reviewer welcome, my JS is non-existent).

[martinpitt]

Wohoo!