Ceph version alignment 18.2.4 and 18.2.7

L

leeker

New Member
May 21, 2025
4
1
3
Hi All,

Very new here - appreciate any help.
I have installed a 3 node cluster running CEPH.
1 of the nodes is running 18.2.7 and the other 2 are running 18.2.4.

I have run all the updates and upgrades both from GUI and from CLI, but i am unable to align the versions.
The warning i am trying to address is from Ceph > Services (Monitors and Managers) proxX 'other cluster members us a newer version of this service, please upgrade and restart'.
root@prox3:~# ceph -v
ceph version 18.2.7 (4cac8341a72477c60a6f153f3ed344b49870c932) reef (stable)
root@prox3:~# apt install ceph-common
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
ceph-common is already the newest version (18.2.7-pve1).
0 upgraded, 0 newly installed, 0 to remove and 0 not upgraded.
root@prox3:~#

root@prox2:~# ceph -v
ceph version 18.2.4 (2064df84afc61c7e63928121bfdd74c59453c893) reef (stable)
root@prox2:~# apt install ceph-common
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
ceph-common is already the newest version (18.2.4-pve3).
0 upgraded, 0 newly installed, 0 to remove and 0 not upgraded.
root@prox2:~#

root@prox1:~# ceph -v
ceph version 18.2.4 (2064df84afc61c7e63928121bfdd74c59453c893) reef (stable)
root@prox1:~# apt install ceph-common
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
ceph-common is already the newest version (18.2.4-pve3).
0 upgraded, 0 newly installed, 0 to remove and 0 not upgraded.
root@prox1:~#

OSD also state 'Outdated OSDs'


Many thanks
 
Hello,

How exactly are you updating packages via CLI? Which repositories are enabled in each host? Are the same repositories available in all hosts?
 
Last edited:
HEALTH_WARN: There are daemons running an older version of ceph
mon.prox1 mon.prox2 osd.0 osd.1 mgr.prox1 mgr.prox2 are running an older version of ceph: 18.2.4
 
leeker said:
HEALTH_WARN: There are daemons running an older version of ceph
mon.prox1 mon.prox2 osd.0 osd.1 mgr.prox1 mgr.prox2 are running an older version of ceph: 18.2.4
Click to expand...
Hi there!

Have you restarted the services as requested? Don't need to restart the server if you can't now, but at least you need to restart the Ceph services.

After the upgrade, restart the services in your cluster, one by one (manager, monitor and metadata - if you have); then, server by server, access the terminal/shell and restart the OSDs with the command "systemctl restart ceph-osd.target". Wait until your Ceph goes "active+clean" again and go to the next server. When finished restarting OSDs in all the nodes, Ceph will go green again.

EDIT: you can restart the services through the web interface, clicking on the node > Ceph > Monitor or node > Ceph > CephFS; click each one, then restart and wait till it come back with the new version. Same for OSDs, if you prefer restart them one by one.

Best regards.
 
Last edited:
Ceph 18.2.4 is the latest version in the Enterprise repository, 18.2.7 is only available in the no-subscription Ceph repository. Please verify /etc/apt/sources.list.d/ceph.list on all your nodes. It is possible you updated Ceph and then configured the Ceph Enterprise repository? Or that one host had both enabled?

I would recommend to enable the ceph no-subscription repository on all nodes, update only the ceph package, and then remove the no-subscription repository (followed by apt update so that the package metainfo matches the enterprise repository). Downgrading packages is not supported, so in my opinion this would be the cleanest solution.
 
  • Like
Reactions: gurubert, cetjunior and leeker
You must log in or register to reply here.
Top Bottom