Environment
Novell NetWare Cluster Services 1.6
Novell Clustering Services 1.0
Novell Clustering Services 1.01
Novell Clustering Services 1.0
Novell Clustering Services 1.01
Situation
How to remove Clustering Services and the Clustering DS Objects
Resolution
This procedure covers decommissioning an entire NCS cluster, not removing one server from a cluster.
There isn't a single click "remove cluster" utility. You must follow the below steps:
1) Shutdown the cluster ("Cluster down" command).
2) Run ULDNCS.NCF on all the servers to unload the cluster NLMs.
3) Load ConsoleOne, highlight the cluster object and delete all its subordinate objects. This will include but is not limited to all resource templates, all resources, and all cluster enabled volumes. You may select multiple objects and delete them as a block. Include the ADMIN_cluster object.
4) Delete the cluster container object (The object must be empty before it can be deleted step #3).
5) In Clustering 1.0x Load NWCONFIG | Standard Disk Options | Modify partitions and Hot fix | Select the Shared device, and delete the Clustering Services (SBD) disk partition(this can be done anytime the cluster NLMs aren't
running).
In Clustering 1.6 or later use NSSMU to delete the cluster SBD
partition only - nothing else. Then remove the SHARED flag from the clustered device so that volumes on it can be mounted by a non-cluster server.
6) Modify the entry in AUTOEXEC.NCF to remove the LDNCS.NCF entry.
7) Modify the entry in the AUTOEXEC.NCF to remove the LOAD NSS \AUTODEACTIVATEVOLUME=ALL entry.
8) After all this is done, and the server comes up clean on a reboot, you possibly also have to recreate the regular volume object in NSSMU, using ALT-F8.
The NDS tree will still contain the cluster schema definition but no NDS objects related to the cluster. The server will still hold copies of the clustering NLMs and NCFs, but this is normal for any NetWare 6.5 server and they do not need to be deleted.
Note:
Should you experience any problems with this procedure such as missing pool or volumes, DON'T TOUCH ANYTHING MORE and call Novell Support immediately.
Search keywords: uninstall uninstalling removing deinstall deinstalling
There isn't a single click "remove cluster" utility. You must follow the below steps:
1) Shutdown the cluster ("Cluster down" command).
2) Run ULDNCS.NCF on all the servers to unload the cluster NLMs.
3) Load ConsoleOne, highlight the cluster object and delete all its subordinate objects. This will include but is not limited to all resource templates, all resources, and all cluster enabled volumes. You may select multiple objects and delete them as a block. Include the ADMIN_cluster object.
4) Delete the cluster container object (The object must be empty before it can be deleted step #3).
5) In Clustering 1.0x Load NWCONFIG | Standard Disk Options | Modify partitions and Hot fix | Select the Shared device, and delete the Clustering Services (SBD) disk partition
6) Modify the entry in AUTOEXEC.NCF to remove the LDNCS.NCF entry.
7) Modify the entry in the AUTOEXEC.NCF to remove the LOAD NSS \AUTODEACTIVATEVOLUME=ALL entry.
8) After all this is done, and the server comes up clean on a reboot, you possibly also have to recreate the regular volume object in NSSMU, using ALT-F8.
The NDS tree will still contain the cluster schema definition but no NDS objects related to the cluster. The server will still hold copies of the clustering NLMs and NCFs, but this is normal for any NetWare 6.5 server and they do not need to be deleted.
Note:
Should you experience any problems with this procedure such as missing pool or volumes, DON'T TOUCH ANYTHING MORE and call Novell Support immediately.
Search keywords: uninstall uninstalling removing deinstall deinstalling
Additional Information
Formerly known as TID# 10015339
Change Log
Updated step 5 to remove destructive advice which would have ended up with data loss for the customer, deleted steps 8-10 (deletion of NLMs and NCFs) which were unnecessary, and added new step 8
Added the step to remove the SHARED flag, and the warning that this TID covers decommissioning an entire cluster only
Added the step to remove the SHARED flag, and the warning that this TID covers decommissioning an entire cluster only