[SOLVED] Restarting pve-cluster service modify corosync.conf with a typo within

[javih]

While trying to add a new node and remove anohter to our cluster we had note the every time we or pvecm restart the pve-cluster service, corosync.conf file is modified with a typo in the first line. The content of this file before we restart begins with:


logging {

debug: off

to_syslog: yes

}


After restart the service the file is:


:ºlogging {

debug: off

to_syslog: yes

}


Due to this the corosync service fails while restarting. Do is there some temp file or whatever that has this typo inside and, therebefore, restarting the service restore this file to the one with the typo?

Our PVe version in all the nodes (3): is pve-manager/6.0-11/2140ef37 (running kernel: 5.0.21-4-pve)

Regards!

 

[fabian]

While trying to add a new node and remove anohter to our cluster we had note the every time we or pvecm restart the pve-cluster service, corosync.conf file is modified with a typo in the first line. The content of this file before we restart begins with:


logging {

debug: off

to_syslog: yes

}


After restart the service the file is:


:ºlogging {

debug: off

to_syslog: yes

}


Due to this the corosync service fails while restarting. Do is there some temp file or whatever that has this typo inside and, therebefore, restarting the service restore this file to the one with the typo?

Our PVe version in all the nodes (3): is pve-manager/6.0-11/2140ef37 (running kernel: 5.0.21-4-pve)

Regards!

there are two copies of the file:
/etc/pve/corosync.conf (stored on pmxcfs)
/etc/corosync/corosync.conf

whenever you (or PVE) write the former, the latter gets overwritten on all nodes that are currently part of the cluster. probably the copy on pmxcfs has the typo, and you only fixed the local one?

 

[javih]

Thank you Fabian, you hit the nail in the head!!!
Our /etc/pve/corosync.conf in the first node was the one with the typo. Now after restarting the cluster-service everithing it's ok.
Before this, in order to add the third node with a workaround, we did the pvecm add, then manualy change the /etc/corosync/corosync.conf file in all the nodes, an then restart corosync for all nodes. That way we had the third node in place.

Best regards!!
Javier