Rename cluster nodes due to HELO Fqdn

[hiho]

Hi all,

i'm running a HA cluster on the non-subscription repo. I need to change both node fqdns to match the HELO that i actually want. Problem is - as soon as i do that (changed the hostname in /etc/hosts,/etc/hostname, /etc/pmg/cluster.cnf) the rsync job stops doing it's job with an "unexplained error".

I also tried the template-modification approach -> copied the main.cf.in to /etc/pmg/templates/main.cf.in, the following line seems to be responsible for the fqdn

smtpd_banner = $myhostname [% pmg.mail.banner %]

Modifying this line does the job, HOWEVER, the file moves from the active node to the second node with rsync -> i modified it to smtpd_banner = mx01.xxx.at [% pmg.mail.banner %] which gives mx01.xxx.at -> but the second node will also return mx01.xxx.at, while it should return mx02.xxx.at, as the file moves from the active to the passive.

Any idea how to get that working? I really do not want to rebuild the cluster from scratch...

Thanks for your help,

Georg

 

[Stoiko Ivanov]

On a hunch - does rsync also fail, if you connect to the new fqdns via ssh first as root? (to add the new names to the authorized_keys file)?

 

[hiho]

i think the original issue is this


-- Logs begin at Tue 2019-05-14 14:42:57 CEST, end at Tue 2019-05-14 14:50:09 CEST. --
May 14 14:43:44 mx01 systemd[1]: Starting Proxmox Mail Gateway Cluster Tunnel Daemon...
May 14 14:43:47 mx01 pmgtunnel[783]: starting server
May 14 14:43:47 mx01 pmgtunnel[783]: starting tunnel 786 10.1.1.60
May 14 14:43:47 mx01 systemd[1]: Started Proxmox Mail Gateway Cluster Tunnel Daemon.
May 14 14:43:47 mx01 pmgtunnel[783]: tunnel finished 786 10.1.1.60
May 14 14:44:06 mx01 pmgtunnel[783]: restarting crashed tunnel 856 10.1.1.60
May 14 14:44:06 mx01 pmgtunnel[783]: tunnel finished 856 10.1.1.60
May 14 14:44:25 mx01 pmgtunnel[783]: restarting crashed tunnel 862 10.1.1.60
May 14 14:44:25 mx01 pmgtunnel[783]: tunnel finished 862 10.1.1.60
May 14 14:44:44 mx01 pmgtunnel[783]: restarting crashed tunnel 875 10.1.1.60
May 14 14:44:59 mx01 pmgtunnel[783]: tunnel finished 875 10.1.1.60
May 14 14:45:13 mx01 pmgtunnel[783]: restarting crashed tunnel 879 10.1.1.60
May 14 14:45:13 mx01 pmgtunnel[783]: tunnel finished 879 10.1.1.60
May 14 14:45:32 mx01 pmgtunnel[783]: restarting crashed tunnel 886 10.1.1.60
May 14 14:45:32 mx01 pmgtunnel[783]: tunnel finished 886 10.1.1.60
May 14 14:46:41 mx01 pmgtunnel[783]: restarting crashed tunnel 1166 10.1.1.60
the pmgtunnel seems to have some issue when the names are changed.

 

[hiho]

ssh'ing from one node to another works just fine.

 

[Stoiko Ivanov]

also with the new name?
did you restart the pmgtunnel service?

 

[hiho]

yep, i can ssh from one machine into another with the new name. i rebooted both nodes as well.

May 14 15:15:52 mx02 systemd[1]: Started Proxmox Mail Gateway Cluster Tunnel Daemon.
May 14 15:15:52 mx02 pmgtunnel[801]: tunnel finished 802 10.1.1.59
May 14 15:16:11 mx02 pmgtunnel[801]: restarting crashed tunnel 884 10.1.1.59
May 14 15:16:11 mx02 pmgtunnel[801]: tunnel finished 884 10.1.1.59
May 14 15:16:30 mx02 pmgtunnel[801]: restarting crashed tunnel 888 10.1.1.59
May 14 15:16:30 mx02 pmgtunnel[801]: tunnel finished 888 10.1.1.59
May 14 15:16:49 mx02 pmgtunnel[801]: restarting crashed tunnel 1061 10.1.1.59
May 14 15:16:49 mx02 pmgtunnel[801]: tunnel finished 1061 10.1.1.59
May 14 15:17:08 mx02 pmgtunnel[801]: restarting crashed tunnel 1223 10.1.1.59
May 14 15:17:08 mx02 pmgtunnel[801]: tunnel finished 1223 10.1.1.59
May 14 15:17:27 mx02 pmgtunnel[801]: restarting crashed tunnel 1391 10.1.1.59
May 14 15:17:27 mx02 pmgtunnel[801]: tunnel finished 1391 10.1.1.59
May 14 15:18:36 mx02 pmgtunnel[801]: restarting crashed tunnel 1871 10.1.1.59
May 14 15:18:36 mx02 pmgtunnel[801]: tunnel finished 1871 10.1.1.59

 

[Stoiko Ivanov]

please post:
* your pmg/cluster.conf
* the output of 'uname -n' on all nodes
* do the node have each other resolvable via '/etc/host'?

Do you try to have both nodes use the same hostname/fqdn?