Currently our users don't know what to expect the failure between config changes should be when considering the following factors
Have cman print the expected failure detection time based upon heuristics.
The goal here is to have CMAN print out a value after taking in to account all of the various timings on the system.
These timings include (but are not entirely limited to; see other bug):
* multipath timeout
* quorum disk timeout
* cman (totem) timeout
This is done in conjunction with execution of heuristics as part of bug 614128 so setting a dependency.
Effectively, this feature is simply designed to print out a message stating how long, in total, the cluster should take to detect a host has died and initiate recovery.
Currently, most of the timers required are printed in the logs, with no indication of what they actually mean.
So, for example, one might see something like the following in the logs as a consequence of implementing this feature:
corosync: [CMAN ] CMAN X.Y.Z (built Dec 24 2012 14:08:21) started
corosync: [CMAN ] Failure recovery initiation time 11.2 seconds
If, you then went over and pulled the plug on a host in the cluster, it should be approximately 11.2 seconds before you see the following message:
corosync: [QUORUM] Members: 1
corosync: [TOTEM ] A processor joined or left the membership and a new membership was formed.
*** Bug 614128 has been marked as a duplicate of this bug. ***
Most of the tuning and documentation has been superseded by work here: