Unexpected Fibre Channel HW path & device assignment

[mvpel]

Hi all,

I've added a batch of recycled disks to a system with dual ds2405 arrays and TL fibre channel cards.

However, one of the 10 added disks has popped up with a strange hardware path.

All of the existing disks, and nine of the new disks, are on hardware paths like so:

0/0/4/0/0.8.0.255.0..0 - upper array
.... c0t_d0
0/0/4/0/0.8.0.255.1..0 - lower array
.... c1t_d0

1/0/4/0/0.8.0.255.0..0 - alt path
.... c3t_d0
1/0/4/0/0.8.0.255.1..0 - alt path
.... c4t_d0

But one of the new disks, which has HP03 firmware just like many of the other disks, is showing up like so:

0/0/4/0/0.8.0.255.7.11.0 - /dev/dsk/c8t11d0
1/0/4/0/0.8.0.255.8.11.0 - /dev/dsk/c7t11d0

The fcmsutil is showing that it has loop IDs of 119 and 123, respectively.

When I move the disk to a different slot, it stays on this odd hardware path, just with a different loop id number corresponding to the new array slot: 0.8.0.255.7.7.0, etc.

Naturally, there's new "fcpdev" devices in ioscan corresponding to the 0.8.0.255.7 path on both adapters.

Does anyone know why this disk would have turned up on this odd path while all the others were attached at the expected locations?

I can replace the disk with another one, but before I go through all that paperwork I want to be sure I won't run into the same problem again.

Thanks for any information you can offer on why this is happening and any suggestions on how to get it back to normal.

[Torsten.]

Are the JBODs daisy chained with different settings on this address switch? Maybe some are using soft addressing?

[mvpel]

Yes - the upper array is Enclosure ID 0, and the lower is Enclosure ID 1, and both Tachyon interface cards can access both arrays.

The enclosure controllers have firmware HP01, so maybe that's something I should look at updating.

How would I tell if they're using soft addressing?

[mvpel]

One other bit that might be of use - the STM is not able to do a SCSI READ CAPACITY on the 0.8.0.255.12.0 disk, so it looks like I've got at least one bad disk in the batch.

[mvpel]

I'm going to try updating the ds2405 firmware from HP01 to HP05 and see if that makes any difference.

[mvpel]

Ok, it appears that this disk was bad, but in a subtle and peculiar way.

I popped in two more disks, to replace the dead one at 12 and this one, and they both attached as expected to the proper hardware path.

Plus, when I pulled up STM information for the oddball hardware path, it showed the latest firmware, but thousands of "non-media" errors. I couldn't see much more detail than that, but I think that's sufficient evidence to pull this disk and send it to the shredder.

So I guess I'll chalk up the odd hardware path to some sort of malfunction in the disk's interaction with the FC enclosure or controller, and move along to more important things.

[mvpel]

.