Notifier Cluster service group type

[Rajeev_Sinha]

Hi,

What is the best recommended way of configuring the notifier cluster service group. Whether the parallel attribute should be 1 or 0.

example below:

[root@node1 ~]# hagrp -display ClusterService |grep -i parallel
ClusterService Parallel              global     0        <--------------------------------------------------0 or 1
[root@node1 ~]#
 

For what I can understand is if we have few application resources running in dirtributed node then to get the notification from all nodes the Cluster service group should be parallel i.e 1.

Appreciate your help.

Thanks and Regards,

Rajeev

[arangari · Accepted solution]

it should be 0. The group should be failover. The ClusterService service group is failover group has it may host WAC resource and GCOIP resource, which are failover by design. The same group can host NotifyMngr resource.  

VCS policy has 'RSM' - replicated state machine, thus every node will generate the same notifications for any event happened on any node. So you do not need to have this group as parallel.  Even if you decide to keep the resource failover, you may get duplicate entries in your email box/snmp trap console, which could be waste and confusing.

the notifications are preserved whenever the notifier process faults (upto a configurable queue size). Whenever the process is restarted (as result restart on same node or failover on other node), the notifications queued-in are processed.

HTH