Hi Marc,
I can share what we did a few months ago. As a remark, I am not sure
Nautilus is available in EL8 but may be I missed it. In our case we did
the following travel:
- Pacific to Octopus on EL7, traditionally managed
- Conversion of the cluster to a cephadm cluster as it makes every
upgrade much more easier and decouple Ceph from the underlying OS
- Reinstallation of all nodes to EL8 : cephadm makes easy to restore the
node after reinstallation as a cluster member, without any reconfiguration
- Upgrade to Pacific with EL8
The key point in our upgrade strategy was the move to cephadm in
Octopus. I think that nothing prevents you from moving from EL7 to EL9
once you have converted your cluster to cephadm (a very straightforward
migration), as Podman on EL9 should work with Octopus as it is the case
in EL9.
Cheers,
Michel
Le 19/04/2023 à 16:14, Marc a écrit :
Sorry for addressing this again. But I think there are quite a few still with Nautilus, that are planning such upgrade.
Nautilus is currently available for el7, el8
Octopus is currently available for el7, el8
Pacific is currently available for el8, el9
Quincy is currently available for el8, el9
el7 eol 2024
I would like to end up at el9 with a the Pacific release that does not have these performance issues[1]
However with the currently offered ceph releases I am more or less forced to an upgrade path of:
upgrade hosts from el7 -> el8, upgrade from nautilus -> pacific, upgrade el8 -> el9
- disadvantage upgrading hosts with some co-located services to el8 and then again to el9
- disadvantage having to wait with el9 upgrade until pacific performance issue is fixed.
the alternative I have been looking at is building Nautilus for el9 (centos9 stream) which seems to go ok (site-test-packages is not building though). This would result in a more efficient upgrade of:
upgrade hosts from el7 -> el9, upgrade from nautilus -> pacific
- advantage is only 1 os upgrade.
- advantage is I can stay on Nautilus until the pacifc performance issue is fixed.
Up until now I can't remember having to much issues with building things from source. However with this storage I would like to have more reassurance.
1. can anyone share experiences with building for different os?
2. what is with all these build warnings, is this common?
3. is it possible that @ceph they can build these rpms and have this go through their default (automated) testing/deploy procedures? I assume with such nice testing facilities this should not take that much time, not?
4. Can anyone recommend some testing procedures? I have a 3 node test cluster. I was thinking of upgrading one host to el9 and run these el9 Nautilus rpms on it, see how it holds.
ps. I am not really interested yet in adding complexity with podman.
pps. Would be nice if someone from ceph could advice on this.
[1]
https://www.mail-archive.com/ceph-users@xxxxxxx/msg19012.html
_______________________________________________
ceph-users mailing list -- ceph-users@xxxxxxx
To unsubscribe send an email to ceph-users-leave@xxxxxxx
_______________________________________________
ceph-users mailing list -- ceph-users@xxxxxxx
To unsubscribe send an email to ceph-users-leave@xxxxxxx
