Environment
Novell Cluster Services
Novell Open Enterprise Server 2 (OES 2) Linux
Situation
The nodes of a working Novell Cluster with POSIX -like ext2, ext3, xfs or reiser- clustered volumes are unrecoverable, unrepairable or deleted (in for instance a virtual environment), but the shared storage is unaffected.
A new cluster is installed and configured and the old cluster resources need to be recreated using the old cluster's shared volumes.
Resolution
For Novell Storage Services (NSS) volumes this is documented in TID 7006969
For POSIX volumes the procedure is a bit different.
Preparation:
The first step is to enable admin_mode for EVMS. This is accomplished by editing /etc/evms.conf.
01. Locate the following line in this config file:
# admin_mode = yes
02. Remove the #
03. Save the file.
Re-enabling the the shared container and volume for Novell Cluster Services:01. Launch evmsgui as root.
02. Verify that the server can see the container from the POSIX resource of the previous cluster under the "Containers" tab in evmsgui. This should still be the same name as used for the previous cluster.
03. Under the "Containers" tab, check which disk device is used by the container by expanding the [container] then Objects Consumed, up to the disk.
04. Right Click the disk device and choose "Remove Segment Manager from Object...". Verify that the "Logical Disk" is the correct one, then click the "Remove" button.
05. Click "Ok" (This will not affect the data). Click the "Save" button, then "Save" again to commit the change.
06. Go to the "Disks" tab and select the logical disk from which you just removed the Segment Manager.
07. Using the "Actions" menu, choose "Add"> "Segment Manager to Storage Object..."
08. Choose "Cluster Segment Manager" and then click "Next".
09. Verify that the Storage object is the disk used in the previous steps, then click "Next".
10. Verify that "Node" is the current node, and that "Type" is set to "private", then fill out the "Name" field using a new or the same name as used previously, then Click "Ok". Click the "Save" button, followed by "Save" to commit the change.
11. Quit evmsgui, not doing so will prevent NCS and NSS from working properly.
12. Edit /etc/evms.conf to undo the change made in the preparation steps.
12. Edit /etc/evms.conf to undo the change made in the preparation steps.
Recreate the Cluster resource:
Using iManager you can now recreate the Cluster resource using the existing Posix Volume the same as described in "Cluster-Enabling a Linux POSIX Volume" in the documentation.
Basically go trough these steps:
01. In iManger, select "Clusters"> "Cluster Options", then browse select the cluster on which you want to recreate the posix cluster resource using the looking glass button.
02. Click "New" under "Cluster Objects"
03. Select "Resource", then click [ Next >> ]
04. Fill out the "Cluster Resource Name: ", for the "Inherit From Template: " field, use the looking glass button to browse towards and select the "Generic_FS_Template".
05. Leave the "Online Resource after Create" unchecked and "Define Additional Properties" ticked, then click [ Next >> ]
05. Leave the "Online Resource after Create" unchecked and "Define Additional Properties" ticked, then click [ Next >> ]
06. Adjust the resources load script appropriately so it uses the container and volume handled in the previous steps, then click [ Next >> ].
07. Adjust the resources unload script appropriately, then click [ Next >> ].
08. Adjust the monitor script, if required and appropriate, then click [ Next >> ].
09. Adjust the Resource Policies, if required, then click [ Next >> ].
10. Adjust the Resource Preferred Nodes if applicable, then click [ Finish ]
After this you have recreated the base cluster resource of a posix file system.
Using the "Creating a Virtual Server Object for the Linux POSIX Volume Cluster Resource" of the documentation this resource can be altered to have a ncp server object and volume export.