[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux-ha
Subject:    [Linux-HA] initdead
From:       Stephen Souness <sounie () hotmail ! com>
Date:       2006-10-24 2:05:21
Message-ID: ehjs3s$jtg$1 () sea ! gmane ! org
[Download RAW message or body]

Hi.

I have been doing a little experimentation to try to come up with some 
suitable values for the various timeouts for heartbeat and the resources 
in my cluster.

To try out the effect of different initdead values in ha.cf I have 
stopped heartbeat from running on both of the nodes in the cluster, and 
then started it on the one that is configured as the primary node.

What I would have expected to see was heartbeat starting up and then 
waiting for the duration specified for initdead, then realising that the 
other node in the cluster was not responsive, then starting up the 
resource(s).

What I have observed is that the delay appears to be 50% longer than the 
initdead that has been specified.

E.g.
- when I set initdead to 120 seconds, my heartbeat log has a time 
difference of around 188 seconds from when heartbeat first starts up and 
when it starts to consider its resources;
- when I set initdead to 240 seconds, my heartbeat log has a time 
difference of around 368 seconds from when heartbeat first starts up and 
when it starts to consider its resources;

I have recorded values of 3 observations for each initdead value being 
set at 60, 120, 180 and 240 seconds and have seen this linear pattern 
emerge quite clearly.

Have I missed something obvious regarding how initdead works?

--
Stephen

_______________________________________________
Linux-HA mailing list
Linux-HA@lists.linux-ha.org
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
[prev in list] [next in list] [prev in thread] [next in thread]