MSCS Cluster Disks

[ozrussells]

Hi Everyone,

I am hoping someone can shed some light on this issue I have been scratching my head about all day.

I am in the process of setting up a two node MSCS cluster across two hosts with fibre attached RDM LUNs. I have followed the most recent guide for setting up a MSCS cluster and attaching the raw disks but have I have got stuck when adding one of the disks to the second node.

I have added an RDM to node 1 correctly, physical mapping and physical SCSI controller on a new SCSI ID 1:0

I then go to the second node and try to add an existing disk and browse to the same VMDK file I just created on node 1 but I am unable to see it.

Even stranger I have a total of 5 disks to add to this cluster, the first time I added all 5 disks at once to node 1 and then tried to add them to node 2 using add existing disk. 4 of the disks listed but this one that I am having problems with was missing. This is what started off this whole thing this morning. The only difference between this problem disk and the other 4 is that this disk is 1.2TB and the rest are below the 1TB mark.

So since then I have been trying to add just this one disk using the guide and searching forums and posts with no luck.

I finally thought that being Vsphere 5 I might be able to add the RDM twice as in add in new disk, raw device mapped disk and then select the LUN. Every bit of literature told me that after it has been added once I should not see it on node 2, but low and behold I can see the LUN and I have added it to each node separately, there is a a vmdk file in each of the guest machines folder on the datastore and I can see the disk in Storage manage in Windows in each guest?

I am sure this is not right but I am very confused at this point.

Any help would be greatly appreciated.

Thanks

Simon

[a_p_]

Welcome to the Community,

Do you see the "missing" .vmdk mapping .vmdk file in the datastore browser? Please use RVTools to check the RDM/.vmdk mappings. This gives you an overview where the files are located and how the mappings look like.

André

[SantoshKumar36]

Hi - Assuming you will be adding physical RDM to the node 1 and node 2.

You can try with another way of doing it. Here is the KB article #http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=101051...

If you add the below settings in the vCenter advanced setting, the LUNs which you have already added to node 1 will be re-displayed at the time , when you add the  new  storage in node 2 and then select the same LUN which you have added for node 1. So both the OS in node 1 and node 2 will have the visibility.

1. In the vSphere Client, select Administration > vCenter Server Settings > Advanced Settings.
2. Depending on your needs, add one of the following or both key and value pairs:
   • config.vpxd.filter.rdmFilter; false  ---> This option is case sentisitve

[ozrussells]

Hey guys,

Thanks for taking time to help me out!

I will definitely give your suggestion a try Santosh and come back with feedback, also thanks Andre.

I can see the RDM VMDK file that is missing in storage browser and also it appears on node 1 in Windows as a disk.

Since posting earlier i did some more investigation. I migrated node 2 to another host, also a different host from node 1 and still cant see this missing disk. I then migrated it to the same host and it popped up straight away when adding it! But i would rather the two nodes sit on different hosts really. It is still weirdthat i can see identically mapped disks from node 1 and 2 if they are smaller then the missing disk, i don't think this is a size issue though being as one node works fine and windows sees the lun. I have also mapped the missing disk to node 2 first then tried from node 1 but same thing happens in reverse.

Will try the suggestion from Santosh soon

Thanks for the help guys

Simon

[ozrussells]

Reporting in

I gave Santosh's suggestion a try but with no luck.

Now for the good news, I have managed to resolve the issue.

I basically started from scratch, detached the LUN, removed the mappings from the SAN, recreated a new mapping on the SAN with a new LUN number, re-scanned on all ESX hosts and then attempted the exact same process to add the RDM to both nodes.

Who knows but this time it worked first time.

Thanks very much for the help chaps.

Cheers

Simon