Wrong scheduler paths and failover problems

Hi all,

I had PBS_HOME in a path in a shared area. Due to a technical problem today, while the cluster was running I had to change the mount point. To do this I

  • changed settings (including pbs.conf) on the secondary
  • rebooted
  • stopped the service on the primary and let secondary take control
  • changed settings (including pbs.conf) on the primary
  • rebooted

this worked well: in both directions the fail over worked. There was one problem: the parameters sched_priv_path and sched_log_path were still the old ones and being it the default scheduler it was not possible to change them (E. C. 15223). I have now resolved with a temporary work around but it’s rather ugly.

I wonder what would be a simple yet stable solution. From what I understand unless I want to set up a multisched configuration I can only wait for a stop then reinstall after a snapshot.

What are your opinions?

thank you