8.2 Removing all GPFS file systems and configuration

 < Day Day Up > 



8.2 Removing all GPFS file systems and configuration

In a test environment, it is often useful to remove all GPFS configuration to the point where it is identical to a fresh install. This is also required if you want to uninstall GFFS.

The broad procedure is as follows:

  1. Unmount all the file systems with umount.

    It is easiest to do this using the dsh command. If you have multiple file systems, you must umount all of them.

  2. Delete all file systems with mmdelfs.

    Again, you must run the mmdelfs command once for each file system but it, like all the remaining commands in this process, need only be run on the GPFS primary node.

  3. Delete all NSDs with mmdelnsd.

    You will most certainly have multiple NSDs. Each one must be deleted using the mmdelnsd command.

  4. Shutdown GPFS on all nodes with mmshutdown -a.

    If you have multiple nodesets, you must run the mmshutdown -a command for each nodeset as follows:

     # mmshutdown -C <nodeset> -a 

  5. Remove all nodes from nodeset with mmdelnode -a.

    The removal should be done for each nodeset in the GPFS cluster using the mmdelnode command as follows:

     # mmdelnode -C <nodeset> -a 

  6. And finally, remove the GPFS cluster definition with mmdelcluster -a.

Example 8-12 shows the commands we used to delete our small GPFS configuration. The commands were run exclusively on the GPFS primary node. All file systems were unmounted before we started.

Example 8-12: Removing all GPFS configuration from the primary node

start example
 [root@storage001 root]# mmdelfs /dev/gpfs0 mmdelfs: Marking the disks as available All data on following disks of gpfs0 will be destroyed:     gpfs1nsd     gpfs2nsd     gpfs3nsd Completed deletion of file system /dev/gpfs0. mmdelfs: Propagating the changes to all affected nodes. This is an asynchronous process. [root@storage001 root]# mmdelnsd 'gpfs1nsd;gpfs2nsd;gpfs3nsd' mmdelnsd: Propagating the changes to all affected nodes. [root@storage001 root]# mmshutdown -a Mon Nov 25 17:54:18 CDT 2002: mmshutdown: Starting force unmount of GPFS file systems Mon Nov 25 17:54:23 CDT 2002: mmshutdown: Shutting down GPFS daemons storage001-myri0.cluster.com:  0513-044 The mmfs Subsystem was requested to stop. node001-myri0.cluster.com:  0513-044 The mmfs Subsystem was requested to stop. node002-myri0.cluster.com:  0513-044 The mmfs Subsystem was requested to stop. node003-myri0.cluster.com:  0513-044 The mmfs Subsystem was requested to stop. node004-myri0.cluster.com:  0513-044 The mmfs Subsystem was requested to stop. storage001-myri0.cluster.com:  Shutting down! storage001-myri0.cluster.com:  'shutdown' command about to kill process 2642 node001-myri0.cluster.com:  Shutting down! node001-myri0.cluster.com:  'shutdown' command about to kill process 18448 node002-myri0.cluster.com:  Shutting down! node002-myri0.cluster.com:  'shutdown' command about to kill process 30090 node003-myri0.cluster.com:  Shutting down! node003-myri0.cluster.com:  'shutdown' command about to kill process 16736 node004-myri0.cluster.com:  Shutting down! node004-myri0.cluster.com:  'shutdown' command about to kill process 29837 Mon Nov 25 17:54:25 CDT 2002: mmshutdown: Finished [root@storage001 root]# mmdelnode -a Verifying GPFS is stopped on all affected nodes ... mmdelnode: Removing old nodeset information from the deleted nodes. This is an asynchronous process. [root@storage001 root]# mmdelcluster -a mmdelcluster: Propagating the changes to all affected nodes. This is an asynchronous process. mmdelcluster: Command successfully completed [root@storage001 root]# 
end example

If you desire, GPFS can be uninstalled at this point. If you are removing GPFS but leaving CSM on the cluster, you should remove all gpfs RPMs, plus rsct.basic. This is easily accomplished by using the smsupdatenode node from the management node as follows:

 # smsupdatenode -ae gpfs.base gpfs.msg.en_US gpfs.docs rsct.basic 

If you no longer have CSM on your cluster, you should remove all GPFS and RSCT RPMs, plus the SRC RPM. You will need to run the following manually on all nodes:

 # rpm -e gpfs.base gpfs.msg.en_US gpfs.docs rsct.basic rsct.core.utils \ rsct.core src 



 < Day Day Up > 



Linux Clustering with CSM and GPFS
Linux Clustering With Csm and Gpfs
ISBN: 073849870X
EAN: 2147483647
Year: 2003
Pages: 123
Authors: IBM Redbooks

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net