lvsplit error code 1 (URGENT!)

[Ola_5]

Can't seem to find a list of error codes. When running lvsplit in backupscript I get error code 1, but when I run lvsplit manually everything works fine!
It stopped working after firmeare upgrading FC10 disk array controllers and disk firmware.

[Paula J Frazer-Campbell]

Hi

If it works manually and does not work from a script then there is an environmental variable problem.

In your script put the full path name to all commands, i.e.

/usr/sbin/lvsplit

and not lvolsplit.

Also ensure that path is correct in your script.

It would also be helpful if you could post your script.


HTH

Paula

[Bill McNAMARA_1]

you should see something in the syslog.log and/or dmesg, but in either case, you should have _some_ output other than a return code.. are you redirecting your errors?

lvsplit blah lvol 2>/tmp/err

Later,
Bill

[Paula J Frazer-Campbell]

Hi Again

I run an FC10 on an N class 11.00 and this script works OK.


date >> ~/var/adm/n0cronlog
################################
# Flush memory
################################
sync
sync
################################
echo "Splitting disks"
lvsplit -s _copy /dev/avro/avro
################################
echo "FSCK on split disk"
################################
fsck /dev/avro/avro_copy
################################
echo "Making mount dir"
################################
mkdir /avrobackup
################################
echo "Mounting split disk to /avrobackup"
################################
mount /dev/avro/avro_copy /avrobackup
################################
echo "RM on /avroTWO dir"
################################
rm /avroTWO/COPYCOMPLETE
rm -R /avroTWO/*
################################
echo "CD to /avrobackup"
################################
cd /avrobackup
################################
echo "CD to /avrobackup and copy to avroTWO"
################################
find .|cpio -pudvm /avroTWO
################################
# Move clear of the mount disk
################################
cd /avro
################################
echo "Unmounting the split disk"
################################
umount /dev/avro/avro_copy
################################
echo "Removing the mount dir"
################################
rmdir /avrobackup
################################
echo "Merging the disk back online"
################################
lvmerge /dev/avro/avro_copy /dev/avro/avro
################################
echo "Avrocopy to avroTWO complete"
###############################
date >> ~/var/adm/n0cronlog
mailx -s "Mirror backup log" paula@avro.co.uk

HTH

Paula

[Ola_5]

Thank's for alla the answers, but the thing is : backupscript worked before firmwareupgrading.
No error messages at all in syslog or dmesg. Environment variables are correct...

[Paula J Frazer-Campbell]

Hi

Can you post your script?

Can you also post your command line that works ?


Paula

[Ola_5]

The problem is not related to the backupscript. I have run the command manually, exactly as in my script, and the attachement is output from the command, and some checks I made.

[Carlos Fernandez Riera]

The file shows that one disk to not belog to this volume group :


Physical volume "/dev/dsk/c9t0d0" does not belong to volume group ....

So, as long /etc/lvmpvg and LVM commands are showing it as configured for this volume group, maybe you must run a vgcfgrestore for that disk.

I think that disk have its LVM headers not updated, due to a hardware or software issue.

Before do vgcfgrestore you can try this:

To see VGID:



vgexport -s -m /tmp/vgap -p /dev/vgxx

or:

VGID is recorded in each disk:

dd if=/dev/rdsk/cxtxtx bs=1024 skip=8 count=9 | dd bs=8 skip=2 count=1 | od -x | head -1 .



running dd for all disks of vg should show the same

See:
http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0x19c7cf38d6bdd5118ff10090279cd0f9,00.html

Check also the lastest patch for LVM.

Regrads.

[Ola_5]

Problem solved!
We have FC10 with fibre disks, and some of the primary disks went offline for a short moment when reprogramming firmware, so the alternative link became primary path, and due to that, the order i /etc/lvmpvg did'nt correspond to output from vgdisplay -v conserning primary/alternative link.
Solved problem temporarely by changing order in /etc/lvmpvg, but soon I will change back the primary links to correct disks.

[Carlos Fernandez Riera]

I realize you are a newcomer...

Please see also: http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0x022718276953d61190040090279cd0f9,00.html


Enjoy the forum.

[Bill Hassell]

Hints on script writing:

Always start *every* script with the courtesy loader string:

#!/usr/bin/sh

That way, the system will always use the same shell to run the script. Without this directive in line 1, the current shell will try to run it.

When a script fails, whether run manually or by cron, run the script with the trace option as in:

sh -x MSycriptName options...

And just for reliability, place an explicit $PATH after the first line:

PATH=/sbin:/usr/sbin:/usr/bin:/usr/lbin

That way, any problems (or security issues) with the current $PATH are eliminated. This statement in the script will not affect your current environment and often fixes problems with cron and other batch jobs.