How to upgrade in production environment

[dpecile]

Hi, I am new to proxmox and have some doubts.
Will be greatly appreciated if someone can help me.

I have installed Proxmox 6.2.12 with ceph in 3 nodes, with HA.
I was not able to found any post or documentation showing how to upgrade with no down time.
Is this possible ? Which es the procedure ?
I want to upgrade to the last ceph version if it's possible, because I am having some slowness in ceph.

Thanks and regards.

Demian

 

[RokaKen]

I use this process:

=================================================================================
# Node maintenance

  
ceph status

    # set cluster in maintenance mode with :

ceph osd set noout


# for node 1..N ; do

    #  migrate VMs and CTs off node

(GUI or CLI) Bulk Migrate

    # run updates

apt update && pveupgrade

reboot  # if required, e.g., kernel update

    # wait for node to come back on line and quorate
    
    # wait for all PGs to recover (active+clean)

    # next N

# restore

ceph osd unset noout

ceph status

# done


==================================================================

 

[dpecile]

Hi RokaKen
Thanks for answer.
So, the idea is to move all the VM & CT to another cluster, upgrade the node, and bring online again, and the other 2 nodes with old versions should connect to ceph ?
I mean it's possible to upgrade 1 node each time ?
Or I need to shutdown/backup everything, and restore from backup after upgrade all nodes ?
Sorry, I am a total ignorant with the ceph storage.

Demian

 

[RokaKen]

Hi RokaKen
Thanks for answer.
So, the idea is to move all the VM & CT to another cluster, upgrade the node, and bring online again, and the other 2 nodes with old versions should connect to ceph ?

You're welcome. Yes, but technically, you are moving all the VMs & CTs to another node within the _same_ cluster. Always migrate from older version node to newer version node in a mixed cluster. The MGR/MON/MDS will elect a new leader and maintain quorum. CEPH version mismatches may produce warnings but will function while the upgrade is in process -- always follow the documentation, e.g. Nautilus to Octopus when upgrading MAJOR versions.

I mean it's possible to upgrade 1 node each time ?

Yes.

Or I need to shutdown/backup everything, and restore from backup after upgrade all nodes ?

That is an alternative -- it's your choice.

Sorry, I am a total ignorant with the ceph storage.

Demian

Regardless of advice you receive in this forum, there is no substitute for the Ceph documentation.

 

[dpecile]

Thanks I will try tonight.
Sure, always I can dig in the docs, and I am reading about.
Let you know, thanks again for the advise.

Demian