Ceph monitor address unknown

Kukwiak

New Member
Dec 12, 2023
9
3
3
Hello,
Today I added another node to the cluster. Before attaching it, I installed ceph. Ceph is version 17.2.6, while on the other nodes it is version 17.2.5. On the new server, I can't enable ceph monitor, even though when I try to do so, I get a response that everything is ok. The message that appears with the "new" monitor is that the address is unknown.
Could the problem be the ceph version difference?
How to debug well, where is the source of the problem?
 
Although I am on 18.2.0 I can say: I am with you. Probably the small difference in the version number is not relevant.

My situation is this: the installation of an additional node went smooth. One OSD (the fourth in my Ceph cluster) is working as expected. There was a "mon" installed automatically, iirc. Now I still have my three "old" (but uptodate) Mons, running and shown with Address.

This new fourth node is shown in the webgui as "Name=mon.xyz - Host=xyz - Status=stopped - Address:Unknown - Version:<empty> - Quorum:No"

Several attempts to delete this thing failed. /etc/pve/ceph.conf lists the stable three nodes while the fourth one is completely missing.

Good luck
Disclaimer: this is a Homelab and I am a newbie in regard to Ceph
 
In my case, the following steps helped:
Disabling and stopping the faulty monitor:
Code:
systemctl disable ceph-mon@hostname
systemctl stop ceph-mon@hostname
Then removing its directory:
Code:
mv /var/lib/ceph/mon/ceph-mon ~
After these steps, the monitor disappeared from the WEB interface and I was able to add it again without any problem.
 
In my case, the following steps helped:
Disabling and stopping the faulty monitor:
Code:
systemctl disable ceph-mon@hostname
systemctl stop ceph-mon@hostname
Then removing its directory:
Code:
mv /var/lib/ceph/mon/ceph-mon ~
After these steps, the monitor disappeared from the WEB interface and I was able to add it again without any problem.
Thanks it work!