Problem: Re-Join of a Server Stalls

Problem

A re-join of an UCS system stalls.

Environment

The system has been joined successfully earlier. It is a mail server with Dovecot app installed and the mail folders (/var/spool/dovecot) are located on a NFS share which is mounted to this mail server.
The join.log the last messages seen are these:

13.05.20 12:46:19.934  LISTENER    ( WARN    ) : finished initializing module nscd_update with rv=0
13.05.20 12:46:19.934  LISTENER    ( WARN    ) : initializing module dovecot-shared-folder
Waiting for data... (interrupt to abort)

In /var/log/kernel.log you might see:

May 14 12:24:59 mail kernel: [86061.078114] lockd: server 10.22.44.55 not responding, still trying

Solution

  1. Use TCP instead of UDP to mount the NFS share. It is much more reliable. Edit /etc/fstab accordingly and match the configuration to your NFS server. Try to use “soft” instead of “hard”.
  2. Tell Dovecot about NFS (see Dovecot Wiki) and our documentation:
    ucr set mail/dovecot/process/dotlock_use_excl=no
  3. Run the join again.
Mastodon