IBM SG24-5131-00 Manual Do Utilizador

Página de 240
Cluster Management and Administration 
173
Note that you cannot add nodes to the resource group list with the DARE 
Resource Migration utility. This task is performed through SMIT.
Stopping Resource Groups
If the location field of a migration contains the keyword 
stop
 instead of an 
actual nodename, the DARE Resource Migration utility attempts to stop the 
resource group, which includes taking down any service label, unmounting 
file systems, and so on. You should typically supplement the keyword 
stop 
with the migration type 
sticky
 to indicate that the resource stays down, even 
if you reboot the cluster.
As with sticky locations, sticky stop requests are superseded by new sticky 
migration requests for the same resource group, or they are removed by 
default,
 non-sticky migration requests for the same resource group. Thus, a 
stopped resource will be restarted at the time of the next migration request.
8.5.3.4  Using the clfindres Command
To help you locate resources placed on a specific node, the DARE Resource 
Migration utility includes a command, 
clfindres
, that makes a best-guess 
estimate (within the domain of current HACMP configuration policies) of the 
state and location of specified resource groups. It also indicates whether a 
resource group has a sticky location, and it identifies that location.
See Appendix A of the 
HACMP for AIX, Version 4.3: Administration Guide, 
SC23-4279, for the syntax and typical output of the 
clfindres
 command. 
8.5.3.5  Removing Sticky Markers When the Cluster is Down
Sticky location markers are stored in the HACMPresource class in the 
HACMP ODM and are a persistent cluster attribute. While the cluster is up, 
you can only remove these locations by performing a subsequent non-sticky 
migration on the same resource group, using the 
default
 special location 
keyword or specifying no location. 
Be careful when using a non-sticky stop request, since the resource group 
will likely be restarted at the next major cluster event. As a result, all 
non-sticky requests produce warning messages. A non-sticky stop could 
be used to halt a cascading resource group that has 
INACTIVE_TAKEOVER set to false during periods in which its primary 
node is down.
Note