note that RHEL 8 is still present in qa/distros/supported-random-distro$/rhel_latest.yaml. this should be removed too, but nobody has added RHEL 9 yet

removed rhel8 after clt discussion

i found that several suites still have symlinks to these old distro files:

qa/suites/fs/verify/distro/rhel_8.yaml
qa/suites/fs/mixed-clients/distro/rhel_8.yaml
qa/suites/fs/workload/0-rhel_8.yaml

i'll go through and change any that aren't upgrade tests to use the {distro}_latest.yaml instead

several of the upgrade suites had broken symlinks to qa/distros/supported/*.yaml, so i replaced those with links under qa/distros/all/

• What about things like qa/distros/single-container-host.yaml which specify:
  os_type: centos
  os_version: "8.stream"

my only objective here is to stop building/testing ceph packages for these old distros. i'll leave any decisions about container host distros to folks that maintain a cephadm suite

• What is the difference between qa/distros/supported and qa/distros/supported-all-distro? If none, it would be nice to ditch one of them while at it.

thanks, i added a commit to replace qa/distros/supported-all-distro with qa/distros/supported

Rados approved: https://tracker.ceph.com/projects/rados/wiki/MAIN#httpstrellocomcPuCOnhYL1841-wip-yuri5-testing-2023-10-02-1105-old-wip-yuri5-testing-2023-09-27-0959

However, @cbodley I do see some centos 8 stream jobs that were scheduled in the run: https://pulpito.ceph.com/yuriw-2023-10-02_20:49:32-rados-wip-yuri5-testing-2023-10-02-1105-distro-default-smithi/

Can you confirm whether the results here worked as intended? If so, feel free to merge.

@yuriw don't merge this; let Casey respond to the above comment ^

@cbodley it scheduled centos 8 stream, as opposed to centos 8 which I see you removed. I think it did work as expected.

• What about things like qa/distros/single-container-host.yaml which specify:
  os_type: centos
  os_version: "8.stream"

@ljflores it scheduled lots of rados/cephadm jobs using centos8 and rhel as the container host, but i chose not to touch container hosts as discussed above. i also see some rados/thrash-old-clients jobs that pulled in qa/distros/container-hosts/centos_8.stream_container_tools.yaml

but now that i look at those logs, i see that even the cephadm jobs are installing ceph packages. when we stop building packages for these old distros, those jobs will start to fail

@cbodley I see; so does that mean this PR needs further modifications?

@cbodley I see; so does that mean this PR needs further modifications?

@ljflores i'd rather follow up on container host stuff in a separate pr. i'm willing to switch centos 8->9 in those yaml files for testing, but would need someone else to debug and fix any issues that uncovers

@ljflores @cbodley can this be merged?

i've opened #53901 to hopefully address the issue with container hosts