[Linux-cluster] How to set node timeout when using gulm?

Lon Hohberger lhh at redhat.com
Fri Feb 1 16:06:43 UTC 2008


On Thu, 2008-01-31 at 17:20 -0800, Vectorz Sigma wrote:
> I'm aware of how to do this for CMAN but I'm running gulm.  I can't
> find information anywhere on how to do this.
> 
> Anyone know?

According to the gulm.5 man page:

       heartbeat_rate
              The rate at which the heartbeats are checked by the
              server  in seconds.  Two-thirds of this time is the
              rate at which the heartbeats are sent.  Default  is
              15.

       allowed_misses
              How  many  consecutive heartbeats can be missed be-
              fore we mark the node expired.  Default is 2.

Remember that allowed_misses+1 is the "actual" node death time.  So,
with the defaults, that's 45 seconds - not 30.  (important!)

<cluster>
   ...
   <gulm heartbeat_rate="10" allowed_misses="5" />  
          <!-- (5+1)*10 second failure recovery -->
   ...
</cluster>

-- Lon




More information about the Linux-cluster mailing list