hi,
do you use ceph within the cluster and how many nodes does the cluster have?
apt full-upgrade should work, but might remove unused packages.
(never use a simple
apt upgrade - this won't upgrade kernel packages and the environment packages like ceph, zfs, qemu ... are heavily depending on this)
The reference
https://pve.proxmox.com/pve-docs/pve-admin-guide.html#system_software_updates recommends using
Code:
# apt-get update
# apt-get dist-upgrade
or you can use the wrappers:
Code:
pveupdate
pveupgrade
# For details, if you are interested:
root@pve:~# whereis pveupgrade
pveupgrade: /usr/bin/pveupgrade /usr/share/man/man1/pveupgrade.1.gz
root@pve:~# strings /usr/bin/pveupgrade
#!/usr/bin/perl
use strict;
use warnings;
use File::stat ();
use Getopt::Long;
use PVE::Tools;
use PVE::Cluster;
use PVE::SafeSyslog;
use PVE::INotify;
....
To upgrade from 8 to 9, changes to the repository files are mandatory. So no worry to accidentially upgrade to far.
(If you ever come to the position check issues in advance with
pve8to9 --full )
In advance to the upgrade you should decide, which services should be migrated in advance to a reboot for availability reasons e.g. by a bulkmigration.
If you manage your VMs anyways, via the PVE HA groups, take a look at the datacenter configuration:
Cluster Configuration > Options > HA Settings > select one of the options
This determs the behavior of the HA enabled guests during a planned shutdown (
https://pve.proxmox.com/wiki/Manual:_datacenter.cfg)
Most likely either freeze or failover will do a good job for you.
Do the upgrade and reboot process server by server, to detect problems early.
After you rebooted the first server, migrate the guest workload back on that server to detect if your workload might have issues with updated virtualization libraries.
Also check if the ceph daemons are healthy after the updates (if in use) before you continue to the next host.
BR, Lucas
