root disk gone faulty

[shikhar_1]

Hi All,

I am facing problem with one of HP-UX box 11.11 whose root disk gone faulty and even not showing in ioscan also.

Server could not able to query with root disk./dev/dsk/c0t6d0 (0/0/0/3/0.6.0)

#-> strings /etc/lvmtab | more
/dev/vg00
/dev/dsk/c0t6d0
/dev/dsk/c0t5d0

#-> setboot
Primary bootpath : 0/0/0/3/0.6.0
Alternate bootpath : 0/0/0/3/0.5.0 (/dev/dsk/c0t5d0)

#-> echo "boot_string/S" | adb -k /stand/vmunix /dev/kmem
boot_string:
boot_string:    disk(0/0/0/3/0.5.0.0.0.0.0;0)/stand/vmunix


#-> lvlnboot -v
lvlnboot: Couldn't query physical volume "/dev/dsk/c0t6d0":
The specified path does not correspond to physical volume attached to
this volume group

[root@cpcwmid5:/.root/sverma8]#

#->
#-> lvdisplay -v /dev/vg00/lvol3 | more
lvdisplay: Warning: couldn't query physical volume "/dev/dsk/c0t6d0":
The specified path does not correspond to physical volume attached to
this volume group
lvdisplay: Warning: couldn't query all of the physical volumes.
--- Logical volumes ---
LV Name                     /dev/vg00/lvol3
VG Name                     /dev/vg00
LV Permission               read/write
LV Status                   available/stale
Mirror copies               1
Consistency Recovery        MWC
Schedule                    parallel
LV Size (Mbytes)            512
Current LE                  32
Allocated PE                64
Stripes                     0
Stripe Size (Kbytes)        0
Bad block                   off
Allocation                  strict/contiguous
IO Timeout (Seconds)        default

   --- Distribution of logical volume ---
   PV Name            LE on PV  PE on PV
   /dev/dsk/c0t5d0    32        32

   --- Logical extents ---
   LE    PV1                PE1   Status 1 PV2                PE2   Status 2
   00000 ???                02032 stale    /dev/dsk/c0t5d0    02032 current
   00001 ???                02033 stale    /dev/dsk/c0t5d0    02033 current
   00002 ???                02034 stale    /dev/dsk/c0t5d0    02034 current
   00003 ???                02035 stale    /dev/dsk/c0t5d0    02035 current
   00004 ???                02036 stale    /dev/dsk/c0t5d0    02036 current
   00005 ???                02037 stale    /dev/dsk/c0t5d0    02037 current
   00006 ???                02038 stale    /dev/dsk/c0t5d0    02038 current
   00007 ???                02039 stale    /dev/dsk/c0t5d0    02039 current
   00008 ???                02040 stale    /dev/dsk/c0t5d0    02040 current
   00009 ???                02041 stale    /dev/dsk/c0t5d0    02041 current
   00010 ???                02042 stale    /dev/dsk/c0t5d0    02042 current
   00011 ???                02043 stale    /dev/dsk/c0t5d0    02043 current
   00012 ???                02044 stale    /dev/dsk/c0t5d0    02044 current
   00013 ???                02045 stale    /dev/dsk/c0t5d0    02045 current
   00014 ???                02046 stale    /dev/dsk/c0t5d0    02046 current
   00015 ???                02047 stale    /dev/dsk/c0t5d0    02047 current
   00016 ???                02048 stale    /dev/dsk/c0t5d0    02048 current
   00017 ???                02049 stale    /dev/dsk/c0t5d0    02049 current
   00018 ???                02050 stale    /dev/dsk/c0t5d0    02050 current
   00019 ???                02051 stale    /dev/dsk/c0t5d0    02051 current

#-> ioscan -kH 0/0/0/3/0
H/W Path       Class                      Description
=====================================================
0/0/0/3/0                   ext_bus       SCSI C1010 Ultra160 Wide LVD A6793-60001
0/0/0/3/0.5                    target
0/0/0/3/0.5.0                     disk    HP 73.4GST373454LC
0/0/0/3/0.7                    target
0/0/0/3/0.7.0                     ctl     Initiator

[root@cpcwmid5:/.root/sverma8]#

My query is that whats the procedure to recity this issue as BDRA information is also seems to be corrupted?

Is that server up with mirrored disk ?

[Matti_Kurkela · Accepted solution]

Your lvdisplay output says "Mirror copies 1" (that is, original + 1 copy), so at least /dev/vg00/lvol3 is mirrored. 

This is confirmed by the fact that Allocated PE = 2 * Current LE, i.e. each logical extent takes 2 physical extents. (In a non-mirrored case, 1 physical extent = 1 logical extent.)

You should check all the other LVs on vg00 and verify that all of them are mirrored... but if you have not yet lost any filesystems in the vg00 volume group, I guess all of them are mirrored.

The lvdisplay -v shows "???" in the PV1 column simply because the system cannot find the matching PV any more (because the /dev/dsk/c0t6d0 is totally dead).

Replacing a mirrored root disk is standard sysadmin procedure on HP-UX. If the failed disk is in a hot-swap slot, you can even replace it without shutting down the server. It just means you must use "lvreduce -k -m 0 /dev/vg00/lvolN" and "vgreduce -f vg00" to remove the failed mirrors if you need to follow the old procedure (delete the failed LV mirrors, remove failed disk from vg00, replace disk, mirror the root disk again).

But if your HP-UX 11.11 is up to date with patches, you can use "pvchange -a N /dev/dsk/c0t6d0" to disable access to the failed disk. Then you can use an easier procedure: replace disk, vgcfgrestore, mkboot, pvchange -a y, vgsync.

HP has a whitepaper titled "When Good Disks go Bad". It contains very good step-by-step instructions for replacing a failed root disk. Use Google to find that whitepaper.

[Nighwish]

Hi

Matti says everything .

Read "When Good Disks go Bad" very good white paper and see what is your case.

Regards.