Typically you would just freeze the group to prevent VCS from acting upon
the failures and ignore the notifications. This is the best course of action.

If you really want to prevent notification, you have to temporarily disable
monitoring. This is done by disabling the resource(s) that will be under
maintenance. You cannot disable a resource unless it is in a clean offline
state on all nodes in SystemList. All the resources depending upon it must
also be in offline state.

Easiest:
A) If its okay to not monitor any resources on the given node for duration
of maintenance, you can completely stop the VCS engine on that node with
"hastop -local -force". The "-force" flag leaves the applications running,
and gets VCS completely out of the picture. Note that other nodes will mark
the group as AutoDisabled until this node rejoins the Port h membership.
A group is AutoDisabled if VCS is unable to monitor all visible nodes in
SystemList, which inhibits VCS from attempting to online the group elsewhere
(an important safety feature).

More involved, but safer:
B) Offline the group, freeze the group, disable resources:

hagrp -offline myGroup -sys myActiveNode
hagrp -freeze myGroup
hagrp -disableresources myGroup
# import DG, perform maintenance
# deport DG
hagrp -enableresources myGroup
hagrp -unfreeze myGroup
hagrp -online myGroup -sys myActiveNode

It is important to let VCS import the DG and start your applications, because
VCS sets noautoimport flag for safety because the DG is visible from several
hosts. When you re-enable the resources, VCS will find the DG imported but
detect the flag isn't set, so it will fault the DG. See "-t import" option
of vxdg manpage for more info.


"Shaun Fogleman" wrote:
>
>I currently have NotifierMngr configured and it works great. But, for database
>maintenance, the database is brought down outside of VCS control which causes
>the NotifierMngr to send its notification of a FAULTED resource. My notification
>levels are ERROR. I want to prevent notification during maintenance windows.
> I tried setting the MessagesQueue to 0 (didn't work). I tried offlining
>the NotifierMngr resource, but the message still gets queued with the Queue
>set to 0. It seems my only choice is to modify the notification levels
to
>SevereError instead of Error during the maintenance.
>
>Any other suggestions?
>
>Thanks!