On Wed, 2009-04-22 at 09:27 -0400, Alan Conway wrote:
> qpidd (my CPG app) currently will loop forever around cpg_init if it returns 
> TRY_AGAIN. This means e.g. that if openais/corosync is not running, it will 
> wait 
> until it is started and then proceed.
> 
> This is good in that qpidd works regardless of whether it is started before 
> or 
> after openais. However it creates confusion among new users trying to get 
> things 
> working for the first time. It seems like qpid should time out, but I'm not 
> sure 
> what a reasonable default timeout is. In particular I want this to play well 
> when openais is started by cman, which seems to have some pretty long 
> timeouts.
> 
> What do other applications do?

cman waits no more than 120 seconds to be fully operation and part of
the cluster and it's a user configurable value via the init script or
via cman_tool -t $timeout value.

None of the cluster daemons, that use cpg, do retry cpg_init if that
fails because all of them connects to openais/corosync before cpg_init
for other reasons. So it is almost certain that they get to cpg_init,
corosync/openais is already running and a failure at that point is
fatal.

I think a 60 seconds timeout is perfectly reasonable timeout value.

Fabio

_______________________________________________
Openais mailing list
[email protected]
https://lists.linux-foundation.org/mailman/listinfo/openais

Reply via email to