Procedure Required to recover the server from splitted LVs

[HP'ian]

Hi All,

I am planning to patch my 9000/800/rp3440, hp 11.11 server. I am planning to do lvsplit before i start the patching. I know the procedure of lvsplit. But i am not sure about how to recover the server if suppose my server is not coming up how can i recover the server from the splitted LVs. Can anyone help me out with the procedure.

[Steven E. Protter]

Shalom,

post this:

lvlnboot -v

read this:
vgdisplay -v /dev/vg00

If it shows boot path to the disks on which the logical volume being split resides, it will definitely boot.

Then you rebuild the mirror with lvextend -m 1 disk name.

Not much of a procedure.

Make sure you have make_net_recovery or make_tape_recovery run in advance so you can recover the system if something unforseen happens.

SEP

[HP'ian]

Hi Steve,

Thanks for your response...My situation is this. After the patching, the server gets rebooted and the server is not coming up. Now my plan is to use the LVs names /dev/vg00/lvol2b, /dev/vg00/lvol3b, etc ,.,which i've created before the patch using lvsplit. Can't i get into lvm mode and rename the dev files lvol2b to lvol2 and old lvol2 to someother name and then boo the system. Won;t the server come up.

Please correct me if am wrong.

Awaiting for your replies.

[Johnson Punniyalingam]

In this case, Hope your boot disks are mirrored ? (both Pri & Sec disks)

if not mirror, better to mirror your secondary disk. Once mirror your secondary disk done

try to boot the server from (Secondary mirror disk)

On the Patching Day

make_tape_recovery via tape/Ignite Server

I would break the mirror, (lvextend -m 0)

deattach the primarydisk

and try patching on Second disk and bring up Server and Check Application (etc...) all works fine,

Than plug-back the primary disk and mirror back again to Secondary disk

[Kapil Jha]

Hi,

Your method should work [renaming lv] in maintenance mode, although I neber tried it...but technically its seems good.

Please make sure you have ignite backup as well for worst scenario.

Do it and let us know if that worked ;)
BR,
Kapil+

[R.K. #]

Hi HP'ian,

For STEP BY STEP process, have a look on SK Chan's response in the below thread:

http://forums11.itrc.hp.com/service/forums/questionanswer.do?threadId=70246

Hope this will clear most of the doubts.

[HP'ian]

Hi All,

First of all, my THANKS to all for your timely valuable suggestions. The following procedure was followed by me to before starting the patching activity.

1.) Split the mirror copies of all the vg00 LVs and run fsck on the backup LVs. Make sure U run lvsplit for /dev/vg00/lvol3 at the last.

2.) Mount /dev/vg00/lvol3b to a temporary mountpoint and edit the fstab with the vg00 LVs as "b"
Mkdir /temp
Mount /dev/vg00/lvol3b /temp
Cd /temp/etc/
cp â p fstab fstab_date

Vi fstab and rename all the vg00 LVs with /dev/vg00/lvol*b
umount /temp

Now the vg00 is made into two copies, each copy will be in each disk.

The above 2 steps were performed to keep my original OS ( vg00 ) safe. Step 2 is performed for the following reason. If my server doesn't comes up after the patching, then i can bring the server from the alternate disk ( in which the patching will not get reflected )and bring back the services up.

I think this is also one of the safest method of patching and also we would not be really dependent on ignite backup.

I've heard that most of thee times, recovering the server from ignite would get failed especially from ignite server. Is it so. In my environment i have all the servers backed up to the ignite server.

[Johnson Punniyalingam]

>>I've heard that most of thee times, recovering the server from ignite would get failed especially from ignite server. Is it so. In my environment i have all the servers backed up to the ignite server.<<

No , true as long as , Your network setup everything is fine, I d'not see any issue

[HP'ian]

Thanks Johnson for your update.

In one of my server, i have the Db filessystems in vg00 itself.

Filesystem
/dev/vg00/lvol3
/dev/vg00/lvol1
/dev/vg00/lvol8
/dev/vg00/lvol7
/dev/vg00/lvol4
/dev/vg00/lvol33
/dev/vg00/lvol32
/dev/vg00/lvol31
/dev/vg00/lvol30
/dev/vg00/lvol_quest
/dev/vg00/lvol11
/dev/vg00/lvol12
/dev/vg00/lvol6
/dev/vg00/lvol41
/dev/vg00/lvol5
/dev/vg00/lvol20

I have only ignite backup running in this server. No other filesystem/Db backup configured. I could also see some errors in the ignite backup sayign that files greater than 8GB are not dumped. Is there any way to over ride this option and backup all the files under vg00.

Awaiting for an Earliest reply.

Below is the part of ignite backup log.


* Archiving contents of blhremdr via tar to
/var/opt/ignite/recovery/arch_mnt/2009-11-02,08:34.
pax: dump/dump/dbf/arsys9.dbf : > 8GB. Not Dumped.
pax: dump/dump/aradmin_exp20071112.dmp : > 8GB. Not Dumped.
WARNING: The pax command returned a non-zero exit status (exit status 1).
* Creation of system archive complete.

WARNING: There was a problem with the list of files/directories to archive.
Input: /var/opt/ignite/recovery/client_mnt/0x0013217C62BD/recovery/2009-11-02,08:34/flist
The following line(s) had too few/incorrect entries:
Problem lines:
line: 17 home/dba/oracle/ 100644 1087212000 0 0 NA 1 0x68 0x40000005
line: 11 opt/ar/db/ 100600 1102082905 0 0 NA 1 0x55a3 0x40000006
* Creating CINDEX Configuration File

* /opt/ignite/bin/manage_index -q -c 2009-11-02,08:34\ Recovery\ Archive
-i /var/opt/ignite/recovery/client_mnt/0x0013217C62BD/CINDEX -u
Recovery\ Archive



======= 11/02/09 11:53:44 GMT make_net_recovery completed with warnings

[Johnson Punniyalingam]

Yes, You can exclude "that mount point" which part db_file system which mentioned by you,

-x exclude=/dev/vg00/xyz

Please do man make_net_recovery --> for more information