DMP disabling paths when a SCSI Inquiry is run

[sathish kannan]

Hello world,

I have a problem with Veritas DMP.

Configuration: HP SD32/11.23/VXDMP,VxVm 3.5 and 4.x/EMC Storage.

Whenever I run, any SCSI inquiry command like "syminq" or "symcfg discover", Veritas DMP disbales all its path to the device at the DMP level. This does not happen to all the devices and my customer tells me that it does not happen on a particuar device either.

The SCSI Inquiry command runs succesfully and list all the devices except the DMP device node "/dev/vx/rdmp/cXtYdZ" value. The device is visble across all paths. IOSCAN shows all the disks as claimed and no errors on syslog related to disk going offline except following message related to DMP node.

Jul 19 16:17:05 sat.com vmunix: NOTICE: vxvm:vxdmp: disabled path 31/0x40000 belonging to the dmpnode 0/0x10
Jul 19 16:17:05 sat.com vmunix: NOTICE: vxvm:vxdmp: disabled path 31/0x80000 belonging to the dmpnode 0/0x10
Jul 19 16:16:55 sat.com syslog: Oracle Cluster Ready Services waiting for Service Guard to start.
Jul 19 16:17:05 sat.com vmunix: NOTICE: vxvm:vxdmp: disabled path 31/0xa0000 belonging to the dmpnode 0/0x10
Jul 19 16:17:05 sat.com vmunix: NOTICE: vxvm:vxdmp: disabled path 31/0xe0000 belonging to the dmpnode 0/0x10
Jul 19 16:17:05 sat.com vmunix: NOTICE: vxvm:vxdmp: disabled path 31/0x180000 belonging to the dmpnode 0/0x10
Jul 19 16:17:05 sat.com vmunix: NOTICE: vxvm:vxdmp: disabled path 31/0x1a0000 belonging to the dmpnode 0/0x10
Jul 19 16:17:05 sat.com vmunix: NOTICE: vxvm:vxdmp: disabled dmpnode 0/0x10
Jul 19 16:17:05 sat.com vmunix: NOTICE: vxvm:vxdmp: disabled path 31/0x80400 belonging to the dmpnode 0/0x50
Jul 19 16:17:05 sat.com vmunix: NOTICE: vxvm:vxdmp: disabled path 31/0xa0400 belonging to the dmpnode 0/0x50
Jul 19 16:17:05 sat.com vmunix: NOTICE: vxvm:vxdmp: disabled path 31/0x1a0400 belonging to the dmpnode 0/0x50
Jul 19 16:17:05 sat.com vmunix: NOTICE: vxvm:vxdmp: disabled dmpnode 0/0x50


526> name=c8t0d4, ap=0, fp=3, tp=3, state=DISABLED
1> subpath=c8t0d4 state=DISABLED
2> subpath=c10t0d4 state=DISABLED
3> subpath=c26t0d4 state=DISABLED


I have tested with DMP 4.1 and 3.5 and looks like problem is happning on both. I believe it as a DMP related issue, either a timing issue or some sort of bug.

Has anyone seen this issue before?

Thansk in advance

Regards
Sathish

[David Child_1]

Sathish,

What type of EMC storage is this (e.g. Clariion, Symmetrix)?

David

[sathish kannan]

David, It's EMC symmetrix DMX-3.
Regards
Sathish

[Sandman!]

Hi Sathish,

Looks like a known bug, check out VxVM 3.5m patches PHCO_34255 and PHKL_34254. Snippet from PHCO_34255 below:
...
...
...
( SR: 8606429739 CR: JAGaf89205 )
Veritas Incident Number: 153191
Symptom:
Inconsistency with ASL, DDL or DMP due to the addition of EMC SYMM(DMX-1000) array to DBAC cluster can result in following DMP disable and enable boot messages on the console:

NOTICE: vxvm:vxdmp: disabled path 31/0xa1100 belonging to the dmpnode 1/0x8

NOTICE: vxvm:vxdmp: disabled path 31/0xb1100 belonging to the dmpnode 1/0x8
...
...
...
http://www2.itrc.hp.com/service/cki/patchDocDisplay.do?patchId=PHCO_34255
http://www2.itrc.hp.com/service/cki/patchDocDisplay.do?patchId=PHKL_34254

~cheers

[sathish kannan]

Hi Sandman,

These two patches are missing. I will install it and let you know the status.

Regards
Sathish

[sathish kannan]

Hi Sandman,

We found the problem.

It's a Veritas DMP device node corruption. For some reason one of the internal disk slice and SAN disk's VXDMP device nodes (/dev/rdmp) were the same. We believe it may be due to some storage reallocation which the customer did by changing SAN components.

Thanks for your asssiatance.

I will be closing the thread.

Regards
Sathish

[sathish kannan]

closing thread.

VXDMP device node corruption