Video Screencast Help

SCSI-3 PR for IO fencing not supported on VMware ESXi Virtual machines?

Created: 04 Sep 2013 • Updated: 06 Sep 2013 | 2 comments
zhangchao's picture
This issue has been solved. See solution.

We have 6 guest OS (redhat 6.3) on 3 physical machine running VMware ESX5.0,each physical machine has 2 VMs.

And SFCFSHA has been installed.

The fencing disks are three 1GB LUN from physical arrays via VMware RDM physical mode.

but vxfentsthdw test shows:

Testing quanziweb01 /dev/vx/rdmp/ibm_ds8x000_0088 quanziweb02 /dev/vx/rdmp/ibm_ds8x000_0088
Evaluate the disk before testing  ........................ No Pre-existing keys
RegisterIgnoreKeys on disk /dev/vx/rdmp/ibm_ds8x000_0088 from node quanziweb01  Passed
Verify registrations for disk /dev/vx/rdmp/ibm_ds8x000_0088 on node quanziweb01  Passed
RegisterIgnoreKeys on disk /dev/vx/rdmp/ibm_ds8x000_0088 from node quanziweb02  Passed
Verify registrations for disk /dev/vx/rdmp/ibm_ds8x000_0088 on node quanziweb02  Passed
Unregister keys on disk /dev/vx/rdmp/ibm_ds8x000_0088 from node quanziweb01  Passed
Verify registrations for disk /dev/vx/rdmp/ibm_ds8x000_0088 on node quanziweb02  Failed
Unregistration test for disk  failed on node quanziweb02.
Unregistration from one node is causing unregistration of keys from the other node.
Disk  is not SCSI-3 compliant on node quanziweb02.
Execute the utility vxfentsthdw again and if failure persists contact
the vendor for support in enabling SCSI-3 persistent reservations

I'm sure that the three 1GB LUN from the disk array support  SCSI-3 persistent reservations IF used in physical machines.

But why they do not support  SCSI-3 PR when used in VMware ESX virtual machines?

Is there any up-to-date official guide for configuring I/O fencing for VCS/CFS on VMware ESX?


Operating Systems:

Comments 2 CommentsJump to latest comment

mikebounds's picture

Yes - see which has attached "CFS and VMware Deployment Guide VMDKs":


UK Symantec Consultant in VCS, GCO, SF, VVR, VxAT on Solaris, AIX, HP-ux, Linux & Windows

If this post has answered your question then please click on "Mark as solution" link below

zhangchao's picture

Thanks Mike.

I have finished the  I/O fencing configuration task using 3 one-node cluster CPS.

The current 《vcs6.0.1 install guide》is somewhat miss-guiding, The Guide suggests to use CPS on SFHA cluster, but it's only suited for large scale enterprise designed to support all production clusters using one set of CPS, because its wasteful, it needs at least 3 SFHA clusters, 6 hosts, 9 IP address.(6 node IPs, plus 3 virtual IPs).

For a small environment, 3 one-node clusters as CPS is enough.

I also find out IF 6 guest OS depolyed on 6 different physical machines, 

The SFCFS cluster via cross-box vmware virtual machines supports SCSI-3 PR for I/O fencing with RDM-P mode disks.