MPT SAS Problem with disk ID

[Wilder Mellotto]

Hi, we have here one rx3600 server running HPUX 11.23 with SAS MPT Adapter.

This SAS adapter contains ths disks:

# sasmgr get_info -D /dev/sasd0 -q lun=all -q lun_locate=all
/dev/rdsk/c1t0d0          0/4/1/0.0.0.0.0           1     5     OFF
/dev/rdsk/c1t2d0          0/4/1/0.0.0.2.0           1     7     OFF
/dev/rdsk/c1t3d0          0/4/1/0.0.0.3.0           1     8     OFF
/dev/rdsk/c1t4d0          0/4/1/0.0.0.4.0           1     4     OFF

The disk in Enc. 1 Bay 4 crashed and the server still running without this disk, because we had a backup VG on it.

When we replaced the failed disk, the MPT SAS adapter gives another ID to the new disk, even in the same Enc.1 Bay.4

# sasmgr get_info -D /dev/sasd0 -q lun=all -q lun_locate=all
/dev/rdsk/c1t0d0          0/4/1/0.0.0.0.0           1     5     OFF
/dev/rdsk/c1t2d0          0/4/1/0.0.0.2.0           1     7     OFF
/dev/rdsk/c1t3d0          0/4/1/0.0.0.3.0           1     8     OFF
/dev/rdsk/c1t6d0          0/4/1/0.0.0.6.0           1     4     OFF

Then, we run this command:

# sasmgr replace_tgt -D /dev/sasd0 -q old_dev=/dev/rdsk/c1t6d0 -q new_tgt_hwpath=0/4/1/0.0.0.4.0

WARNING: This is a DESTRUCTIVE operation.
This might result in failure of current I/O requests.
Do you want to continue ?(y/n) [n]...y
ERROR: Unable to obtain SAS Address for the specified HW Path: No such device


Can anyone help us to solve this problem? We want restore the ID 4 to this new disk.

NOTE: In another server with same problem, the command above solved this issue but in this case, this server was rebooted, and the other server that worked fine, was not rebooted.

Thanks.

[Torsten.]

This command is wrong:

# sasmgr replace_tgt -D /dev/sasd0 -q old_dev=/dev/rdsk/c1t6d0 -q new_tgt_hwpath=0/4/1/0.0.0.4.0


must be

# sasmgr replace_tgt -D /dev/sasd0 -q old_dev=/dev/rdsk/c1t4d0 -q new_tgt_hwpath=0/4/1/0.0.0.6.0


(device file of the OLD disk, hardware path to the NEW disk!)

[Wilder Mellotto]

Ok, I posted with the wrong command, but this is correct and the disk was not renamed.

# sasmgr replace_tgt -D /dev/sasd0 -q old_dev=/dev/rdsk/c1t4d0 -q new_tgt_hwpath=0/4/1/0.0.0.6.0

WARNING: This is a DESTRUCTIVE operation.
This might result in failure of current I/O requests.
Do you want to continue ?(y/n) [n]...
ERROR: Could not stat the file /dev/rdsk/c1t4d0: No such file or directory

[Torsten.]

Please post

# saslist get_info lun sasd

# sasmgr get_info -D /dev/sasd0 -q raid

# ioscan -fn

# vgdisplay -v <vg_in_question>

# ll /dev/dsk

# ll /dev/rdsk

[Wilder Mellotto]

Please see attached file.

Thanks

[Torsten.]

Well, I assume you did something like

# rmsf -k -H 0/4/1/0.0.0.4.0

and for this reason the device file

/dev/dsk/c1t4d0

is now deleted.

I also assume you tried with a new disk, and now with a second new disk.

For this reason you see

target       5  0/4/1/0.0.0.5   tgt            NO_HW       DEVICE

You have 2 choices now.

1) get the old device file back (c1t4d0)

2) remove the old device file c1t4d0 from vg00 and extend vg00 with c1t6d0

Let's try 1) now.

# mknod /dev/dsk/c1t4d0 b 31 0x014000

# mknod /dev/rdsk/c1t4d0 b 188 0x014000

Check file permissions and ownership of the files created!

Now proceed with the "sasmgr" command.

We can try 2) if this does not work.

By the way, replacing a SAS disk in "HBA mode" is a difficult procedure with 11.23, it is getting even more complicated in 11.31!

For this reason I always recommend to use these disks in a hardware RAID mode!