One resource do not stop the others…
I was looking for “that” option in Sun Cluster 3.2, some way to configure a notimportantresource in a importantresourcegroup that did not stop the whole “thing” in the case that notimportantresource fail. Today i could find the answer in the “cluster export” command. I was looking in the xml output when i did see the line:
<parameter name="Failover_mode" extension="false" per-node="false" \\ description="Modifies recovery actions taken when the resource fails." \\ tunability="anyTime" type="enum" enumList="NONE HARD SOFT RESTART\\ _ONLY LOG_ONLY" default="HARD"/>
Take a look at the parameter Failover_mode, looks interesting right? And the values? The default for that resource in the example is HARD, but there are other options like… NONE and… LOG_ONLY!
So, i did try that procedure just adding the option:
… and just work! I did “mess” the start/stop scripts and did try to bring the resourcegroup online. I have got the error:
(C450598) On node node2, resource group important-rg is online but resource notsoimportant-rs failed to start
The resourcegroup status was Online faulted, and the resource status was: Start failed (Faulted).
So, when you have a resource that is not “big deal” if it does not start, you can choose between “NONE” or “LOG_ONLY” for the Failover_mode property. I think LOG_ONLY is fine..