Data directory location

RajeshGawai,

You can move the RS Data Directory to a SAN drive.  However when it is being used on a MS Cluster you must not move the Data Directory to a Shared Disk Drive.  It must appear to the Windows OS as a permanent Drive letter and cannot be a shared disk between cluster nodes.  Also, you must have the same configuration on all the nodes in the cluster.  For example, if you move the Data Directory from the default location to the E:\RepliStorData folder on one  of the cluster nodes, you must have the Data Directoty in the same location on the other nodes of the cluster.  The E: drive cannot be a shared device and all cluster nodes must be configured alike.

One other thing to note is that RS 6.2 is no longer a supported version of RepliStor.  I suggest you upgrade to 6.4, SP1, HF1 as soon as possible in order to have a supported environment.

Thanks for reply, we will plan to upgrade RS 6.2 to 6.4 with sp1.

Is there any documentation for to upgrade process?

Also if possible do share document or link which stating 'Data Directory' should not be located on shared disk. This will help me to prove my point to management.

Thanks in advance.

Thanks & Regards,

Rajesh.

The following is from the RepliStor 6.4 Release Notes P/N 300-008-927 and can be found on Powerlink. I believe it is also in the Install Guide which is also on Powerlink.

RepliStor and Microsoft clustering

Consult the note below before installing or upgrading RepliStor in a

Microsoft clustering environment.

! CAUTION

When installing RepliStor in a Microsoft cluster, you must install

RepliStor in the same directory path on each node. If all nodes are

not installed in the same path, Cluster Server may not be able to

use RepliStor resources on all nodes. In addition, when upgrading

RepliStor installations that are in a Microsoft cluster to this

version, all nodes must be upgraded; otherwise, the RepliStor

resources may not be functional on one or more nodes.

Although the document does not explicitly state that the Data Directory must “not” be located on a shared disk, RepliStor may bluescreen the machine if the Data Directory is moved to a Shared Disk resource. The reason is that the KernelCache.rdf is always in that folder and it contains a list of all the file operations that must be sent to the Target node. The RepliStor device driver actually performs dual writes of the mirroring file operations. One copy is in Kernel Memory and the other is written to this KernelCache.rdf file. RepliStor then references this file when mirroring data to the Target. If you were to detach the drive that contains the Data Directory the Kernel Driver would lose its place and may jump to a non-existent location in memory. If that occurred, then the Kernel memory would be corrupted and the machine would crash. Also, each machine must have its own KernelCache.rdf and in a shared disk configuration only one machine at a time has access to the shared Disk resource.

It must be clear that each machine running RepliStor must have continuous access to the Data Directory on that node. The Data Directory can be moved to a different drive but that drive must not be a shared resource. A shared resource can only be accessed by one node at a time and that would prevent RepliStor from running on any other node that does not have access to that drive continuously. It will certainly bluescreen the machine if RS loses access to its Data Directory.

I hope that is more clear as to how RepliStor must be configured in a MS Cluster environment. Thanks…JS