Forum Discussion

Stephan's avatar
Stephan
Level 2
17 years ago

VCS 5.0 and resource faults

Hello,
 
Is there a way to "delay" a VCS resource from faulting until a certain time period has passed?  To make a long story short, we are trying to get a core dump(as a result of an issue), but VCS faults a resource and the core dump does not fully complete.  Thanks in advance for any tips or ideas
 
Steve
 
PS We are testing Solaris LDOMS, so that is where the core dump issue is coming from.

6 Replies

  • Hello Stephan,
     
    Really the easiest way to do this is by marking the desired resource as not critical during the testing process. As long as the resource or child resources are flagged as critical, they will automatically fail over as configured.
     
    Thanks,
    J
  • J,
     
    I was referring to having a delay in the resource "faulting".  We are not failing over anything in this particular issue(and do not have the resource marked as critical).  Thanks.
     
    Steve
     
  • Stephan,
     
    I apologize for misunderstanding your intent. In this case, you can modify the 'ManageFaults' attribute of the resource. The attribute defaults to ALL, I believe. You can change this setting to NONE and configure VCS to leave the service group running as normal and pretty much ignore the resource fault. It will, however, hang the resource group and all other groups in the cluster by placing the cluster into an ADMIN_WAIT state, requiring intervention to clear the fault manually. When the ManageFaults attribute is set to none, the cluster calls a trigger that you can configure to determine how the cluster responds. If this trigger isn't configred, the default action VCS takes when a resource comes offline unexpectedly is to take the group offline. You can change that behavior by modifying the trigger. The trigger name is 'resadminwait'. Please see pgs 424 and 501 in the VCS Admin Guide for further reference in this configuration.
     
    Hope this helps!
    J
  • Hmm...after looking at the trigger further, I was mistaken. That trigger is for notification only and is non-configurable. Researching further.
     
    J
  • After reading through the ManageFaults attribute, I believe if you switch that attribute to NONE, VCS will leave the resouce and all service groups online, but place the cluster into an ADMIN_WAIT state. That state will need to be cleared manually.
     
    Sorry for the confusion!
     
    J
  • you need to consider FaultOnMonitorTimeout & ToleranceLimit Attribute  values to delay the fault of resource.