VMware Cloud Community
mwiederkehr
Contributor
Contributor

problems with target on Windows Storage Server 2008

Hello

We have set up a target on a Windows Storage Server 2008 to use it as a fallback if our SAN appliance goes down.

Unfortunately we encountered 2 problems:

- Performance is bad, especially latency. Throughput ist OK, but latency is often over 100ms. The storage server is in the same network as the appliance, so I think the network isn't the problem.

- From ESX 3.5 we can move a VM to the storage server without any problem. But from ESX 4.0 U1 we get many of the following log entries and the migration fails:

Mar 11 14:17:31 arizona vmkernel: 28:22:08:02.408 cpu0:4096)ScsiDeviceIO: 747: C

ommand 0x2a to device "naa.60003ff64a982d5fbde3bcfcbe59bee6" failed H:0x0 D:0x2

P:0x0 Valid sense data: 0x5 0x21 0x0.

Mar 11 14:17:31 arizona vmkernel: 28:22:08:02.408 cpu3:4160)J3: 2158: Aborting t

xn 0x4100b79aa330 due to failure pre-committing: Not supported

Mar 11 14:17:31 arizona vmkernel: 28:22:08:02.408 cpu3:4160)BC: 3600: Failed to

flush 32 buffers of size 8192 each for object 'VM_Test

.vmdk' f530 28 3 4b026fb9 941c5289 25001a20 f6c4e0b3 214005c4 86 0 0 0 0 0: Not

supported

Both servers have an QLogic 406Xc HBA.

Strangely enough we can start a VM on ESX 4.0 U1 after we migrated it to the storage server from an ESX 3.5. We can even move it back to the storage appliance with ESX 4!

Do you have any suggestions?

Thank you!

Reply
0 Kudos
3 Replies
J1mbo
Virtuoso
Virtuoso

Somebody else might correct me but I didn't think that MS WSS is supported.

Please award points to any useful answer.

binoche
VMware Employee
VMware Employee

Mar 11 14:17:31 arizona vmkernel: 28:22:08:02.408 cpu0:4096)ScsiDeviceIO: 747: C

ommand 0x2a to device "naa.60003ff64a982d5fbde3bcfcbe59bee6" failed H:0x0 D:0x2

P:0x0 Valid sense data: 0x5 0x21 0x0.

this message means LOGICAL BLOCK ADDRESS OUT OF RANGE, does it say out of drive capacity, can you reduce drive capacity and recheck the result? thanks

binoche, VMware VCP, Cisco CCNA

mwiederkehr
Contributor
Contributor

There is plenty of free space on the LUN. I tested some combinations an found out that it works if I use the software initiator.

ESX 3.5 QLogic => works

ESX 4.0 U1 QLogic => doesn't work

ESX 4.0 U1 SW initiator => works

The firmware of the Qlogic HBA is updated to the newest version available.

Seems to be a problem of the QLogic driver in ESX 4, isn't it?

Reply
0 Kudos