mount /dev/vg07/lvol1 /u06 unrecognized file system? what's vg07?

[Gary L]

I prefer to vgexport /dev/vg07, let c12t1d2 leave vg07. I wanna configure c12t1d4 to the new vg07. Just keep the 40GB in system.

As you know, it's pretty interesting, vgcfgrestore just recovery one lvol1, and 0 size. where is the lvol2?

That server run Oracle DB and anther application. That vg07 be created at 2005, in system no mount point use current vg07 and c12t1d2.

Thanks a lot everybody. I will check more doc, see if I could find the purpose and usage.

Any idea?

[Torsten.]

Last try before removing this VG.

# vgcfgrestore -f /etc/lvmconf/vg07.conf.old -l

# ll /etc/lvmconf

Results?

[Steven E. Protter]

Shalom,

The disk count discrepancy pointed out by Torsten tells me that a disk was there and went away for some reason.

Perhaps a disk on the SAN was withdrawn by the SAN admin.

I think Torsten's last try won't work and you need to remove the volume group.

SEP

[piyush mathiya]

Gary,
According to me, you can use this disk for anothere purpose, just follow this steps to reuse it.

# vgchange -a n vg07
# vgexport /dev/vg07
# pvcreate -f /dev/rdsk/c12t1d2
# cd /dev
# mkdir vg07
# mknod group c 64 0X070000
# vgcreate /dev/vg07 /dev/dsk/c12t1d2
# lvcreate -L /dev/vg07
# newfs -F vxfs -o largefiles /dev/vg07/rlvol1
# mkdir /mnt
# mount /dev/vg07/lvol1 /mnt

#####################################


Regards,
Piyush Mathiya

[Gary L]

# vgcfgrestore -f /etc/lvmconf/vg07.conf.old -l
Volume Group Configuration information in "/etc/lvmconf/vg07.conf.old"
VG Name /dev/vg07
---- Physical volumes : 1 ----
/dev/rdsk/c12t1d2 (Non-bootable)

root@srs083:/> ll /etc/lvmconf
total 39680
---------- 1 root root 0 Aug 14 2004 lvm_lock
-rw------- 1 root sys 322560 Aug 11 16:00 vg00.conf
-rw------- 1 root sys 322560 Aug 1 16:26 vg00.conf.old
-rw-r--r-- 1 root sys 82 Aug 1 16:26 vg00.mapfile
-rw------- 1 root sys 870400 Aug 11 16:01 vg01.conf
-rw------- 1 root sys 870400 Aug 1 16:26 vg01.conf.old
-rw-r--r-- 1 root sys 9 Aug 1 16:26 vg01.mapfile
-rw------- 1 root sys 1091584 Aug 11 16:01 vg02.conf
-rw------- 1 root sys 1091584 Aug 1 16:26 vg02.conf.old
-rw-r--r-- 1 root sys 9 Aug 1 16:26 vg02.mapfile
-rw------- 1 root sys 1093632 Aug 11 16:01 vg03.conf
-rw------- 1 root sys 1093632 Aug 1 16:26 vg03.conf.old
-rw-r--r-- 1 root sys 9 Aug 1 16:26 vg03.mapfile
-rw------- 1 root sys 1093632 Aug 11 16:01 vg04.conf
-rw------- 1 root sys 1093632 Aug 1 16:26 vg04.conf.old
-rw-r--r-- 1 root sys 9 Aug 1 16:26 vg04.mapfile
-rw------- 1 root sys 1091584 Aug 11 16:01 vg05.conf
-rw------- 1 root sys 1091584 Aug 1 16:26 vg05.conf.old
-rw-r--r-- 1 root sys 8 Aug 1 16:26 vg05.mapfile
-rw------- 1 root sys 1091584 Aug 11 16:01 vg06.conf
-rw------- 1 root sys 1091584 Aug 1 16:26 vg06.conf.old
-rw-r--r-- 1 root sys 8 Aug 1 16:26 vg06.mapfile
-rw------- 1 root sys 1091584 Aug 1 16:26 vg07.conf
-rw------- 1 root sys 1091584 Aug 1 16:18 vg07.conf.old
-rw-r--r-- 1 root sys 8 Aug 1 16:26 vg07.mapfile
-rw------- 1 root sys 1091584 Jul 31 17:37 vg08.conf
-rw------- 1 root sys 1091584 Jul 29 16:54 vg08.conf.old
-rw-r--r-- 1 root sys 8 Aug 1 16:26 vg08.mapfile
-rw------- 1 root sys 1091584 Aug 11 16:01 vg09.conf
-rw------- 1 root sys 1091584 Aug 1 16:26 vg09.conf.old
-rw-r--r-- 1 root sys 9 Aug 1 16:26 vg09.mapfile
-r-------- 1 root sys 1091584 Apr 9 2007 vg18.conf
-r-------- 1 root sys 1091584 Apr 9 2007 vg18.conf.old


This server down and could not startup at Aug, 1. I recoveried OS through Ignite-UX net_recovery at Aug, 1

[piyush mathiya]

Gary,
According to me, you can use this disk for anothere purpose, just follow this steps to reuse it.

# vgchange -a n vg07
# vgexport /dev/vg07
# pvcreate -f /dev/rdsk/c12t1d2
# cd /dev
# mkdir vg07
# mknod group c 64 0X070000
# vgcreate /dev/vg07 /dev/dsk/c12t1d2
# lvcreate -L /dev/vg07
# newfs -F vxfs -o largefiles /dev/vg07/rlvol1
# mkdir /mnt
# mount /dev/vg07/lvol1 /mnt

#####################################


Regards,
Piyush Mathiya

[Gary L]

Thanks a lot, SEP!

Thank you vey much for your detail steps, Piyush.

[piyush mathiya]

and check that 0X070000 must not in use, to check this use the below steps.
# cd /dev
# ll */group

then you can check it . . .

[Torsten.]

Your /etc/lvmtab lists vg00..vg08, but you have a config for an vg18 too.
What *was* it?

Maybe someone did not delete previous VGs correctly - this happens very often ...

Maybe one of your very first lines (see top of page):

# vgcfgrestore -n /dev/vg07 /dev/rdsk/c12t1d2

was wrong and restored another config to the disk.


Anyway, *this* vg07 is IMHO lost ...

[piyush mathiya]

Torsten,
We can recover it, but for that we dont know in which server it is used erlier, how many disk used with this volume group and logical volume, it was not swap because it is 40 GB disk, and very less place we use this much of space for swap memory, but actully what is the problem is we must know the background of the disk. then and then we can recover it.
In hp unix with LVM, no one can say we lost the data without making any mistek...

Regards,
Piyush Mathiya

[Gary L]

All above words that I posted here whatever commands and outpout are all copy from system (uging Putty remote access that server) paste directly to here.

So, i'm sure type is correct.

[Steven E. Protter]

Shalom,

Swap should be set to between 1.0 to 1.5 times memory.

If you actually start paging to swap, its time to do a memory upgrade anyway.

I'd use this opportunity to make swap a manageable, normal size within the vg00 volume group.

SEP

[Torsten.]

Question is: was this disk really in vg07? Was it the only one?

I see you still have an (old) mapfile:

vg07.mapfile

in /etc/lvmconf

What is the content?

[Gary L]

Hi SEP

Would you please tell me more about "I'd use this opportunity to make swap a manageable, normal size within the vg00 volume group".

Currentlyl, that server RAM is 16GB, pri swap is 4GB

# lvdisplay /dev/vg00/lvol2
--- Logical volumes ---
LV Name /dev/vg00/lvol2
VG Name /dev/vg00
LV Permission read/write
LV Status available/syncd
Mirror copies 0
Consistency Recovery MWC
Schedule parallel
LV Size (Mbytes) 4096
Current LE 256
Allocated PE 256
Stripes 0
Stripe Size (Kbytes) 0
Bad block off
Allocation strict/contiguous
IO Timeout (Seconds) default

You said "... within the vg00 volume group" Could we recreate or increate size of lvol2 pri swap?

Thanks.

[Gary L]

Hi Torsten

Question is: was this disk really in vg07? Was it the only one?

As you know all the info of vg07 I got from system no pager record.

remove /etc/lvmtab and vgsacn -v

# strings /etc/lvmtab
/dev/vg00
/dev/dsk/c2t0d0
/dev/dsk/c2t1d0
/dev/vg01
/dev/dsk/c12t0d0
/dev/vg02
/dev/dsk/c12t0d1
/dev/vg03
/dev/dsk/c12t0d2
/dev/dsk/c12t0d3
/dev/dsk/c12t0d4
/dev/vg04
/dev/dsk/c12t0d5
/dev/dsk/c12t0d6
/dev/dsk/c12t0d7
/dev/vg09
/dev/dsk/c12t1d0
/dev/vg06
/dev/dsk/c12t1d1
/dev/vg07
/dev/dsk/c12t1d2
/dev/vg05
/dev/dsk/c12t1d3
/dev/vg08
/dev/dsk/c17t0d0

So, I think, just one SAN disk c12t1d2 belong to vg07.

/etc/lvmconf> cat vg07.mapfile
1 lvol1

[Steven E. Protter]

Shalom again,

>>
Would you please tell me more about "I'd use this opportunity to make swap a manageable, normal size within the vg00 volume group".
<<

Yes.

My swap philosophy is a modified version of the A. Clay Stephenson model.

1) Swap should be reserved and the system should be managed to never actually have to page processes to it. If its paging regularly it needs more memory.

2) Swap should be part of vg00 the boot volume group because it allows us to create a self contained Ignite backup of the system for disaster recovery that will work immediately after a restore.

3) swap should be part of the boot volume group. It should not be overly large, and systems with a lot of memory don't need to follow the old twice memory rule. A. Clay sometimes recommended having two swap areas within vg00. 1 small and high priority a second larger one a lower priority in hopes that it will never get used.

4) Having your swap areas spread around can cause I/O contention.

5) Having swap on a SAN means if the SAN goes out your system will suddenly stop.

SEP

[Gary L]

Thanks SEP,

I assign the points first and read your swap suggestion later.

[James R. Ferguson]

Hi Gary:

Regarding swap, here's another viewpoint.

Turn on the kernel's 'swapmem_on'. This allows the kernel to count about 7/8 of your physical memory as swap. For every process started, swap must be reserved regardless of whether or not it is actually used. You never really want to do swap I/O as this degrades already degraded performance. From that standpoint, memory becomes relatively cheap to your business.

One of the simplest monitors is 'vmstat'. If "page-outs' (the 'po' column') begin to rise into double-digit values, you have substantial to severe memory pressure and too much swapping is occuring.

You need primary swap space to boot. Having (for example) primary device swap at only 50% of your memory isn't a problem when you need to boot and get a server running.

You can (and should) add some secondary device swap. I like to assign this to one or more non-vg00 volume groups. You can set the swap priority to a lower ordinal value than that of primary swap thus allowing this secondary swap space to be used in preference to the primary space. If I have to ever do swap I/O, I like to have it occur on other than vg00.

If you choose to assign equal swap priorities to primary and secondary swap spaces, you can interleave the swap I/O between them. That too is potentially useful as long as the primary and the secondary device swap spaces are *not* on vg00. You don't want disk head movement added to your already degraded performance as the primary and secondary swap I/O is round-robined.

How much swap you need is application depenedent. The key issue is process swap reservation. ENOMEM errors and "can't fork" are the primary indicators of insufficient swap. A combination of reasonable device swap in appropriate places, plus 'swapmem_on=1', plus adequate physical memory, plus sound system management makes for a well performing system.

Regards!

...JRF...

[Gary L]

Hi JRF

Thank you very much for your kindly help again.

Have a gooday.

-Gary

[Steven E. Protter]

Shalom,

normal practice is to assign based on the value of the information.

I like JRF's suggestion.

You might want to collect some more performance data.

http://www.hpux.ws/?p=6

vmstat is a good tool and I don't think I ever got around to integrating into the patch set above.

SEP