vgscan errors

[Stephen Keane]

HPUX 9000/785 (B1000) running 11.11
No OnlineJFS.

Suspected lvmtab file was out of date/currupt, so renamed it and ran vgscan

result:

# vgscan
Creating "/etc/lvmtab".
vgscan: Couldn't access the list of physical volumes for volume group "/dev/vg00".

Pid 2930 received a SIGSEGV for stack growth failure.
Possible causes: insufficient memory or swap space,
or stack size exceeded maxssiz.

Pid 2930 received a SIGSEGV for stack growth failure.
Possible causes: insufficient memory or swap space,
or stack size exceeded maxssiz.
Couldn't stat physical volume "/dev/dsk/c0t0d0":
Invalid argument
Physical Volume "/dev/dsk/c3t5d0" contains no LVM information


Notes

VG /dev/vg00 should contain PV /dev/dsk/c3t6d0.

Disk /dev/dsk/c3t5d0 is not a PV and is not in any VG.

Disk /dev/dsk/c0t0d0 is a CD-ROM.

Also 'vgdisplay -v' for /dev/vg00 is

# vgdisplay -v /dev/vg00
--- Volume groups ---
VG Name /dev/vg00
VG Write Access read/write
VG Status available
Max LV 255
Cur LV 10
Open LV 10
Max PV 16
Cur PV 1
Act PV 1
Max PE per PV 2500
VGDA 2
PE Size (Mbytes) 4
Total PE 2168
Alloc PE 1398
Free PE 770
Total PVG 0
Total Spare PVs 0
Total Spare PVs in use 0




--- Physical volumes ---
PV Name /dev/dsk/c3t6d0
PV Status available
Total PE 2168
Free PE 770
Autoswitch On

Filesystems are all up and mounted. bdf shows the following

Filesystem kbytes used avail %used Mounted on
/dev/vg00/lvol3 143360 34126 102470 25% /
/dev/vg00/lvol1 111637 61097 39376 61% /stand
/dev/vg00/lvol8 716800 285969 407380 41% /var
/dev/vg00/lvol7 1630208 917024 668907 58% /usr
/dev/vg00/lvol4 65536 33212 30367 52% /tmp
/dev/vg00/lvol10 1024000 242871 732372 25% /patches
/dev/vg00/lvol6 462848 242092 207007 54% /opt
/dev/vg00/lvol5 1024000 132230 836086 14% /home

swapinfo gives

Kb Kb Kb PCT START/ Kb
TYPE AVAIL USED FREE USED LIMIT RESERVE PRI NAME
dev 524288 0 524288 0% 0 - 1 /dev/vg00/lvol2
reserve - 86380 -86380
memory 175344 33276 142068 19%

Note, this isn't a production system.

[Stephen Keane]

Just noticed that /etc/mnttab is a bit weird too (the last line).

/dev/vg00/lvol3 / vxfs log 0 1 1106302742
/dev/vg00/lvol1 /stand hfs defaults 0 0 1106302743
/dev/vg00/lvol8 /var vxfs delaylog,nodatainlog 0 0 1106302748
/dev/vg00/lvol7 /usr vxfs delaylog,nodatainlog 0 0 1106302748
/dev/vg00/lvol4 /tmp vxfs delaylog,nodatainlog 0 0 1106302749
/dev/vg00/lvol10 /patches vxfs delaylog,nodatainlog,nolargefiles 0 0 1106302749
/dev/vg00/lvol6 /opt vxfs delaylog,nodatainlog 0 0 1106302749
/dev/vg00/lvol5 /home vxfs delaylog,nodatainlog 0 0 1106302749
buffy:(pid911) /net ignore ro,intr,port=662,map=-hosts,indirect,dev=0000 0 0 1106302776


buffy is the server name BTW. But /net is not and has never been a mount point.

[Eric Antunes]

Hi Stephen,

Make a backup of your mnttab file and re-mount all log. volumes:

# mv /etc/mnttab /etc/mnttab.bak
# mount -a

Best Regards,

Eric Antunes

[Stephen Keane]

I've tried that, but it makes no difference. I even edited /etc/mnttab (naughty) and removed the last line, 'mount -a' is happy, the lines is still missing until you do vgscan again. It's vgscan that's appears to be writing the line into /etc/mnttab, but only if you remove /etc/lvmtab first!

[Stephen Keane]

Oh, and PID 911 is

root 911 1 0 10:19:36 ? 0:00 /usr/lib/netsvc/fs/automount/automount -f /etc/auto_master

If that helps! I think there are 2 problems going on here, the vgscan errors and the curious line in /etc/mnttab. Not sure if they are related.

[RAC_1]

You have the last entry in mnttab because, the automounter is in play. So just ignore it. First move back the renamed lvmtab.


Then look at solving errors about stack growth failure etc. dmesg/syslog.log

swapinfo -mat
glance -m

Anil

[Stephen Keane]

Anil,

# swapinfo -mat
.............Mb......Mb......Mb...PCT..START/......Mb
TYPE......AVAIL....USED....FREE..USED...LIMIT.RESERVE..PRI..NAME
dev.........512.......0.....512....0%.......0.......-....1../dev/vg00/lvol2
reserve.......-......84.....-84
memory......171......72......99...42%
total.......683.....156.....527...23%.......-.......0....-

(spaces replaced by '.'s)

I don't have glance installed (as I said, this isn't a production server).

[Stephen Keane]

Nothing in dmesg or syslog. Server reboots happily, filesystems are happy, only vgscan seems unhappy. Problem persists even after reboot.

[Eric Antunes]

Please post the result of:

#ioscan -fnC disk

[RAC_1]

What happens if you do vgscan -pv-a

-p is for pre-view.

[Stephen Keane]

Eric, Anil, see attached txt file

[Eric Antunes]

Maybe you need to reinstall the special files:

#insf -e -C disk

[Stephen Keane]

# insf -e -C disk
insf: Installing special files for sdisk instance 0 address 10/0/14/0.0.0
insf: Installing special files for sdisk instance 1 address 10/0/15/1.5.0
insf: Installing special files for sdisk instance 2 address 10/0/15/1.6.0
# vgscan
vgscan: The physical volume "/dev/dsk/c3t6d0" is already recorded in the "/etc/lvmtab" file.

Pid 4968 received a SIGSEGV for stack growth failure.
Possible causes: insufficient memory or swap space,
or stack size exceeded maxssiz.

Pid 4968 received a SIGSEGV for stack growth failure.
Possible causes: insufficient memory or swap space,
or stack size exceeded maxssiz.
Couldn't stat physical volume "/dev/dsk/c0t0d0":
Invalid argument
Physical Volume "/dev/dsk/c3t5d0" contains no LVM information

[Eric Antunes]

kmtune -lq maxssiz output?

[Stephen Keane]

# kmtune -lq maxssiz
Parameter: maxssiz
Current: 0x800000
Planned: 0x00800000
Default: 0x00800000
Minimum: -
Module: -
Version: -
Dynamic: No

[Eric Antunes]

Hi,

Increase it to 134217728: that's my value. You can do this in SAM in a easy way...

[RAC_1]

From the shell,

ulimit -Sa

If using ksh, then ulimit -s unlimited

If that does not work, then bump the value for maxssiz and maxssiz_64bit a little and try

Anil

[Stephen Keane]

# kmtune -lq maxssiz
Parameter: maxssiz
Current: 134217728
Planned: 134217728
Default: 0x00800000
Minimum: -
Module: -
Version: -
Dynamic: No
# vgscan -va
Creating "/etc/lvmtab".

Pid 2756 received a SIGSEGV for stack growth failure.
Possible causes: insufficient memory or swap space,
or stack size exceeded maxssiz.

Pid 2756 received a SIGSEGV for stack growth failure.
Possible causes: insufficient memory or swap space,
or stack size exceeded maxssiz.
Couldn't stat physical volume "/dev/dsk/c0t0d0":
Invalid argument
Physical Volume "/dev/dsk/c3t5d0" contains no LVM information


/dev/vg00
/dev/dsk/c3t6d0


Scan of Physical Volumes Complete.

[Stephen Keane]

# ulimit -Sa
time(seconds) unlimited
file(blocks) unlimited
data(kbytes) 262144
stack(kbytes) 8192
memory(kbytes) unlimited
coredump(blocks) 4194303
nofiles(descriptors) 60

[RAC_1]

Do as follows

ulimit -s unlimited
and try vgscan. do you get this error for commands also??

Anil

[Stephen Keane]

# ulimit -s unlimited
sh: ulimit: The specified value exceeds the user's allowable limit.

I've not seen the error on any other command. I've used vgdisplay, lvdisplay, pvdisplay, rebuilt the kernel and rebooted.

vgdisplay and vgscan are the same executable (hard linked).

[Stephen Keane]

Just as a matter of interest, does your /etc/lvmconf look similar to this (allowing for differences in the # of VGs etc)

# ll /etc/lvmconf
total 746
---------- 1 root sys 0 Dec 14 11:02 lvm_lock
-rw------- 1 root sys 190464 Jan 21 09:27 vg00.conf
-rw------- 1 root sys 190464 Jan 21 09:25 vg00.conf.old
-rw-r--r-- 1 root sys 64 Jan 14 16:19 vg00.mapfile

[Florian Heigl (new acc)]

Maybe check Your vgscan binary?
(it just sort of smells like that)

[Stephen Keane]

I'be tried both /sbin/vgscan and /usr/sbin/vgscan.

One thing I've just noticed, if you run vgdisplay in a user other than root ...

$vgdisplay
vgdisplay: /etc/lvmconf//lvm_lock: Permission denied

I've tried removing lvm_lock, but vgdisplay only works if you are root (lvm_lock is recreated BTW). More clues?

[Robert-Jan Goossens_1]

Stephen,

Florian could be right.

# which vgscan
/usr/sbin/vgscan
# ll /usr/sbin/vgscan
-r-sr-xr-x 31 root sys
557056 May 18 2000 /usr/sbin/vgscan

Robert-Jan