SAN Controller failover breaks access to vg01

our open-e (v6) is configured with a QLOGIC dual-port f/c hba. Which is connected to a HP MSA 2000 with dual-contollers. On the MSA2000 LUN we created a VG (volume group), which is "shared" over iscsi to some of our hyper-v hosts. In the VG Manager we see two Units with the same serial number, S001 is in state (in use by vg01) the other S002 is available.


During firmware upgrade, a Controller Failover of the LUN from Controller A to Controller B happens. Currently the DSS loses connectionton the LUN and the iSCSI connections are not available until we reboot the DSS.

How can we correct the DSS configuration, so a SAN Controller failover (a-b) is "survived"?:confused:

regards,
Walter Behmann

2010/01/15 18:18:56 journal commit I/O error
2010/01/15 18:35:00 [] kernel_thread_helper+0x7/0x10
2010/01/15 18:35:00 [] kernel_thread_helper+0x7/0x10
2010/01/15 18:24:27 [] kernel_thread_helper+0x7/0x10
2010/01/15 18:24:27 [] kernel_thread_helper+0x7/0x10
2010/01/15 18:19:46 [] kernel_thread_helper+0x7/0x10
2010/01/15 18:19:46 [] kernel_thread_helper+0x7/0x10
2010/01/15 18:35:00 [] scsi_execute_req+0x72/0x80
2010/01/15 18:35:00 [] scsi_execute_req+0x72/0x80
2010/01/15 18:19:46 [] scsi_execute_req+0x72/0x80
2010/01/15 18:19:46 [] scsi_execute_req+0x72/0x80

DSS does not yet support multipathing. This is evident by the fact that you see two disks; one available and one in use.
I'm not sure why this hasn't been implemented yet. It's been stable on every Linux distribution for the longest time and of course, Open-E could limit it to specific devices/vendors.

Yes, multipathing would be nice (with the DSS as the target).