lvmtab problems when removing disk

[Lynne Seamans]

Had one disk in a mirrored pair go south.

un-mirrored the one lvol that was on the disk, but even with vgreduce -f it said there were PE's still in use... even though i really didn't think there were other lvols on there.

used vgreduce -l vg02 /dev/dsk/c1t0d0
to get bad disk out of there. Seemed to work.

BUT:

the VG is active, however i cannot do a vgcfgbkcp on it cause it says lvmtab is out of date with the running kernel, kernel sees two disks.

vgdisplay -v vg02 shows only one.

file system in one lvol is up and running fine.

I'm confused. On one hand, my system is functional, on the other, there's a weird error and i could have a problem lurking.

Any ideas?

[Hasan Atasoy]

hi lyenne;

did you see any pe in the output of

pvdisplay -v /dev/rdsk/c1t0d0

if you cannot see try to recreate lvmtab by

mv /etc/lvmtab /etc/lvmtab.11012008
vgscan -a


hasan

[Lynne Seamans]

Thanks for the tips. Here's what i got:

pvdisplay -v /dev/rdsk/c1t0d0
pvdisplay: Physical volume "/dev/rdsk/c1t0d0" is not a block special file.

then when i ran vgscan -a, YIKES, it looks bad:

r:/etc> vgscan -a
Creating "/etc/lvmtab".
Couldn't stat physical volume "/dev/dsk/c3t2d0":
Invalid argument


vgscan: has no correspoding valid raw device file under /dev/rdsk.
Verification of unique LVM disk id on each disk in the volume group
/dev/vg02 failed.




*** LVMTAB has been created successfully.
*** If PV links are configured in the system.
*** Do the following to resync information on disk.
*** #1. vgchange -a y
*** #2. lvlnboot -R

Now vgdisplay -v vg02 says it's not there!

I'm putting the old lvmtab back :)

[Hasan Atasoy]

what about

pvdisplay -v /dev/dsk/c1t0d0
and
pvdisplay -v other_disk_in vg02

Hasan

[Lynne Seamans]

r:/etc> pvdisplay -v /dev/dsk/c1t0d0
pvdisplay: Couldn't find the volume group to which
physical volume "/dev/dsk/c1t0d0" belongs.
pvdisplay: Cannot display physical volume "/dev/dsk/c1t0d0"

r:/etc> pvdisplay -v /dev/dsk/c2t0d0 | more
--- Physical volumes ---
PV Name /dev/dsk/c2t0d0
VG Name /dev/vg02
PV Status available
Allocatable yes
VGDA 2
Cur LV 1
PE Size (Mbytes) 4
Total PE 4340
Free PE 1340
Allocated PE 3000
Stale PE 0
IO Timeout (Seconds) default
Autoswitch On

--- Distribution of physical volume ---
LV Name LE of LV PE for LV
/dev/vg02/loralogs 3000 3000

--- Physical extents ---
PE Status LV LE
00000 current /dev/vg02/loralogs 00000
00001 current /dev/vg02/loralogs 00001
00002 current /dev/vg02/loralogs 00002
00003 current /dev/vg02/loralogs 00003
00004 current /dev/vg02/loralogs 00004
00005 current /dev/vg02/loralogs 00005
00006 current /dev/vg02/loralogs 00006
00007 current /dev/vg02/loralogs 00007
00008 current /dev/vg02/loralogs 00008
00009 current /dev/vg02/loralogs 00009

(lots more)

[Hasan Atasoy · Accepted solution]

what about

vgreduce -f vg02

Hasan

[Lynne Seamans]

Well, it's my production system, and at least it's running OK now.

(university, students moving back, had database down a long time already :( )

I'm going to try that EARLY monday AM.

I'm thinking that is the absolute right answer, and that i can limp along till a better time.

Thank you thank you. I will post a followup as soon as i'm comfortable running our little experiment.

[Mridul Shrivastava]

First thing.. If you just want to replace the bad disk which is mirrored, it could have been done without reducing the lv and vg. using pvchange -a n
so it will detach the PV from VG and then after replacement execute pvchange -a y to attach it again and restore the vgcfg on the disk.. everything will be as it was earlier.
however since u have already removed the disk from the VG and however kernel information is not updated then vgreduce -f vg02 will remove the disk information from the kernel for the disk which are not present in lvmtab.

So as suggested vgreduce -f should resolve the issue. Please post the outcome after Monday task.