SCSI-3 Persistent Reservation Failed on Clustered VMs

[ouchway]

Hi All. I have three clustered virtual machines running Win2008 as guest OS and using Microsoft Cluster. I have shared physical RDM LUNs mapped to each server. Each LUN is added on all three VMs. The LUNs are visible on all three machines. However, when I run Microsoft cluster validation tests it fails with SCSI-3 Persistent Reservation when it gets to testing cluster disk validity. I've got storage mapped from a NetApp filer. I've searched for solutions but Microsoft seems to point to storage. But on the storage level, everything seems to be working pretty fine. This has gotten our project stuck at a point. Please help. Thanks.

[AndreTheGiant]

Are you using LSI serial controller?

See also:

Andre

[AWo]

We had such an issue with a Linux based SAP cluster system (Steel Eye). We needed to enable SCSI-3 on the SAN front-end port (EMC Symetrix Host Director) and SCSI-3 Persistent Reservation on the LUN layer.

I don't now the syntax for NetApp and how this is managed there, but in our case, that was done in the storage backend.

AWo

VCP 3 & 4

\[:o]===\[o:]

=Would you like to have this posting as a ringtone on your cell phone?=

=Send "Posting" to 911 for only $999999,99!=

[ouchway]

Thanks, dears.

Yeah... I'm using LSI Logic SAS. I've searched for NetApp's config for this and found non online. Also logged a support with NetApp but they're yet to get back to me. Anyone had this same experience working with NetApp?

[ouchway]

Any help? Kindly let me know if you've got any ideas. Tnx.

[AWo]

Asked NetAPP already?

AWo

VCP 3 & 4

\[:o]===\[o:]

=Would you like to have this posting as a ringtone on your cell phone?=

=Send "Posting" to 911 for only $999999,99!=

[ouchway]

I have opened a case and no one is attending to it, though we have support. They're kinda evasive about this. The country Netapp guys are also doing same - pushing it back and forth.

[preiter]

Hi,

we've got the same Problem with an Windows Server 2008 R2 Failover Cluster on ESX 4.0.

But at first I must say sorry for my terrible english because I'd no practice since 2005.

We also had shared physical RDM LUN's on a NetApp Filer with OnTap 7.3.1 or 7.3.4 and we had a long discussion with NetApp because of the validation Check. Yesterday a NetApp Consultant help us to find the correct way to configure the LUN on the Storage and map this LUN as RDM to all three VM's. There a lots of different configurations which can made, but there is a difference in the support of MSCS Cluster.

At first we must distinguish two Configurations: are all three cluster nodes on one physical host or are the virtual machine cluster across physical host?

LUN configuration on NetApp should like this:

/vol/vol_vmsql02/qtsql02/sql05.lun 5.0g (5371107840) (r/w, online, mapped)

Serial#: W-D4ZJ/Reto2

Share: none

Space Reservation: enabled

Multiprotocol Type: windows or windows_2008

Maps: vmware_without_alua=103

Remember ALUA is not supported for RDM's

One physical host:

SCSI Controller must be LSI Logic SAS and it should be another Controller than the vmdk has

SCSI Bus Sharing must be set to Virtual

the RAW Device Compatibility Mode must also be set Virtual

Mount the disk to the other VM's with the option use an existing virtual disk and pick the RDM "vmdk-File" from the Datastore

VM's across hosts: the Configuration is nearly the same but the SCSI Bus sharing must be Physical

SCSI Controller must be LSI Logic SAS and it should be another Controller than the vmdk has

SCSI Bus Sharing must be set to Physical

the RAW Device Compatibility Mode must also be set Virtual

Mount the disk to the other VM's with the option use an existing virtual disk and pick the RDM "vmdk-File" from the Datastore

Both configurations work fine, the SCSI-3 Persistent Reservation check succeded and our cluster is still alive.

if you have an older OnTap Version please update the storage because there would be a lot of errors when the windows cluster is alive.

I hope it helps!