How to recreate cluster resource object

  • 7006969
  • 30-Sep-2010
  • 08-Nov-2012


Novell Cluster Services
Novell Open Enterprise Server 2 (OES 2) Linux Support Pack 2


If you are ever in the situation where you need to recreate a cluster resource object associated with an NSS pool, the following steps should help. 


1. Make sure you backup the cluster resource information so that recreating it with the same information is possible.  For example: Load / Unload / Monitor scripts,  IP Address of the resource, NCP server name, and specific commands added to the scripts.  This will be in a supportconfig output or a cluster report generated through iManager. 
2. Offline the cluster resource.
3. Clean up the remaining cluster resource objects. There are generally four objects associated with a cluster pool resource. 
  a) Cluster resource object inside the cluster container.  
The remaining objects are in the parent container.  
  b) Virtual NCP Server object for the resource. 
  c) Pool object. 
  d) Volume object(s).  
Note: NSSMU will not allow you to update NDS without all the objects being cleaned up first. This is why you would see the 20011 error. Once the objects are cleaned up I generally like to run an NDSrepair or give it 5 minutes to make sure eDirectory has had a chance to clean up.
4. Stop clustering on this node. rcnovell-ncs stop.
5. Start clustering. rcnovell-ncs start
6. Load NSSMU and activate the pool and volume. Use the F8 and F4 keys to update into eDirectory. If clustering isn't loaded on this node and you try do this on a shared device,  you will get the following error(error 20833 = zERR_POOL_SHARED_NO_BROKER).
On OES2 SP2 the NDS Object gets created when manually activating the pool and volume through NSSMU so if you get an error stating the object already exists, this is the reason.
7. Once the pool and volume objects are created on one of your nodes in the cluster,  it's time to cluster enable the pool.
Note:  I also check "cluster resources" command output and see whether the resource still thinks it's there. Sometimes you have to bounce the Master IP resource node to get things to clean up.
8. Now use iManager --> Cluster Options to create a new Cluster Pool where you select the pool object that was just updated into eDirectory, You'll need to add the cluster IP address back in and any specific commands you had in the cluster resource scripts. 
9. Online the new cluster resource and test.