[Linux-cluster] rgmanager service initialization on startup

[Mark Hlawatschek]

Hi,

we internally discussed the behaviour of the rgmanager resource group 
initialization, and we came across the following problem:

The initialization process includes a stop operation for all defined resources 
in the resource group. I.e. every time a cluster member starts the rgmanager 
(e.g. after a reboot) all service agents are called with a stop option, 
although the service is not running on that node. 
To track problems and service transitions, our resource agents drop emails to 
operation everytime a service has been stopped. Now they receive email alerts 
caused from rgmanager startups, although everything is ok and the services 
are still running on another cluster node. As you might imagine, the support 
people would suffer regular heart attacks ;-).

Is it really important to stop unstarted services after a reboot ? If so, 
maybe a service attribute (e.g. <service name="foo" cleanstart="yes"/> could 
be introduced to restain this behaviour. 

What do you think ? 
-- 
Gruss / Regards,

Dipl.-Ing. Mark Hlawatschek
http://www.atix.de/
http://www.open-sharedroot.org/

**
ATIX - Ges. fuer Informationstechnologie und Consulting mbH
Einsteinstr. 10 - 85716 Unterschleissheim - Germany