On Thu, 02 Jul 2009 17:25:02 -0700
Marion Hakanson <[email protected]> wrote:

> [email protected] said:
> > I think you should log a support call with Sun, there's probably going to be
> > more to this issue than what can solve easily via a mailing list. 
> 
> The 2nd-line support engineer thinks it might be bug 6851364, which
> unfortunately shows up as "closed";  There's a related one, 6566270,
> but you need contract access to SunSolve to see it.  They're going to
> research it further and get back to me tomorrow.
> 
> If I interpret the bug description correctly, even though we have supported
> devices, firmware, and drivers, it may be that the drives are not producing
> mpxio-compatible GUID's.
> 
> Thanks for the replies so far, it's nice to know I hadn't missed
> something obvious.


ah... two bugs that I've touched :-)

The second related bug has the synopsis

6566270 Seagate Savvio 10k1 disks do not enumerate under scsi_vhci  

which has some comments that are ok for publication in this thread:

==============================================================================
In order for a device's GUID to be acceptable to MPxIO (thus making the device
enumerable under /scsi_vhci), the GUID must have an Association field value of
00B:

00B     Associated with the addressed logical unit
01B     Associated with the target port that received the request
10B     Associated with the target device which contains the logical unit
11B     Reserved

(from SPC-4, section 7.6.3.1).

If the value is 01B or 10B then we cannot guarantee that the GUID is unique
to a particular logical unit, and therefore scsi_vhci will refuse to enumerate 
it.

Example SCSI INQUIRY(6) page83 responses:

Seagate 73Gb Savvio10k1 SAS disk, model ST973401L:

00 83 00 0c 01 23 00 08 .....#..
50 00 c5 00 00 2c c0 d8 P....,..


Fujitsu 72Gb SAS disk, model MAV2073RC:

00 83 00 3c 01 03 00 08 ...<....
50 00 00 e0 11 46 07 30 P....F.0
61 93 00 08 50 00 00 e0 a...P...
11 46 07 32 61 94 00 04 .F.2a...
00 00 00 01 03 28 00 18 .....(..
6e 61 61 2e 35 30 30 30 naa.5000
30 30 45 30 31 31 34 36 00E01146
30 37 33 30 00 00 00 00 0730....


Note that byte #5 in the Fujitsu case is 0x03, and in the Seagate case
is 0x23.
==============================================================================


The first bug you mentioned
6851364 mpt driver is not configuring multipathing against SATA drives loaded 
on a (J4400).  

was closed for exactly the reason above - the disks were producing
ASCII guids, not binary - and mpt doesn't allow such devices to be
enumerated under scsi_vhci / MPxIO. 

>From your initial post, it seems that you've got those same drives.


James
--
Senior Kernel Software Engineer, Solaris
Sun Microsystems
http://blogs.sun.com/jmcp       http://www.jmcp.homeunix.com/blog
Kernel Conference Australia - http://au.sun.com/sunnews/events/2009/kernel
_______________________________________________
storage-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/storage-discuss

Reply via email to