Invalid values on /etc/lvmtab file

[Faizer]

I created the following volume groups using SAM on rp7410 servers running HPUX11.11. Then in viewing the etc/lvmtab file using strings I get the following abnormal entries.
/dev/vg01
a?%M#
/dev/dsk/c15t0d5
/dev/vg03
a?%R
/dev/dsk/c15t1d0
/dev/dsk/c16t1d0
/dev/dsk/c5t1d0
/dev/dsk/c9t1d0

When I reboot the server these volumes are not mounted.
A vgdisplay says that they are not there.
When I do a vgchange on these volume groups, it allows me to mount my file systems.
Please help me on this issue?
Why does this happen. Any solutions GURU's
Thankyou in Advance
Faizer

[Steven E. Protter]

mv /etc/lvmtab /etc/lvmtab.old

vgchange -a n vg01
vgchange -a y vg01
strings /etc/lvmtab
vgdisplay -v


This should help.

SEP

[A. Clay Stephenson · Accepted solution]

The fundmental problem is that the VG is not automatically activated.

Look in /etc/lvmrc. If AUTO_VG_ACTIVATE=0 then no VG's other than vg00 will be automatically activated. This is normal for a ServiceGuard cluster. If that is the casem then add vg01 to the custom_vg_activation function in /etc/lvmrc.

As for the "abnormal" entries in /etc/lvmtab, remember that this is binary data. The strings command simply looks for sequences or 3 or more printable ASCII characters. If it find such a sequence, it outputs it. You simply found a sequence of binary data which is also printable -- this is perfectly normal and expected for a strings /etc/lvmtab.

[Pete Randall]

Faizer,

First, are you using "strings /etc/lvmtab" to view?

Second, try moving the existing file out of the way and re-creating it:

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



Pete

[Faizer]

Hi Gurus
1. I did a vgchange -a n vg01 and vgchange -a y vg01 and checked the lvmtab file. Still it remains the same.

2. I did a vgscan still its the same

3. the /etc/lvmrc file has auto chnage set to 0. But my other volume groups namely vg02 mounts automatically and there is no garbage as mentioned for this volume group in the /etc/lvmtab file
.
I also have the following entry in the /etc/lvmrc file.
custom_vg_activation()
{
# e.g. /sbin/vgchange -a y -s
/sbin/vgchange -a y vg01
/sbin/vgchange -a y vg02
# parallel_vg_sync "/dev/vg00 /dev/vg01"
# parallel_vg_sync "/dev/vg02 /dev/vg03"

return 0
}
In addition I have these vg's on a va7410 array. Also I have a mC service guard cluster.

Today we created a new lun and assigned a vg for it.
Then I got into this situation.
Thanks for your replies. Waiting for your respsonses.
regards
Faizer

[Faizer]

Hi; Gurus

Steve I find that you have shed some light into this problem.

vg02 got mounted on the reboot because of the entry in the lvmrc file.

The abnormal entry could be attributed to your logic.
I do run a MC service guard environment and the activation is set to 0.

Would this invalid entry in the lvmtab file have any effects on my volume groups etc.
Please advise.
Once again thanks for all you Gurus for helping me on time.
thanks
Faizer

[A. Clay Stephenson]

Again, note that when you manually do a vgchange -a y you can then mount your filesystems. This means that /etc/lvmtab is NOT corrupt. This has to be a problem in your vgchange command in /etc/lvmrc. Check it carefully for syntax errors.

Again, the "abnormal" characters are almost certainly simply an artifact of the way the string commands work. Most of the time these "abnormal" strings do not appear because 3 printable ASCII character sequences are not detected. In your case, you simply got "lucky" and 3 or more characters were printable.