Is it possible to share a logical volume by two versions of HP-UX on the same server?

[Johnson Punniyalingam]

sharing "volume group" except "vg00" can be done in clustering called as "shared volume group"

purpose of shared volume group -> you move "data from shared volume to from One server to another server" if they sharing same storage

man vchange " -s, -c" options will give some ideas.

[johnsonpk]

Johnson>>sharing "volume group" except "vg00" can be done in clustering called as "shared volume group"

No need to have shared volume group and all here .
Since both of your disks are having os images and you are booting from any one of the disk at any given point of time(means ,second disk is unused as far as the os running on the first disk is concerned ) you can mount the lv's on the second disk

(same steps that are used for importing a vg )

1)mkdir /dev/vg00_v3
2)mknod /dev/vg00_v3/group c 64 0x010000
3)vgimport -v vg00_v3
4)mount the lv's on temporary mount points

[Johnson Punniyalingam]

>>No need to have shared volume group and all here .
Since both of your disks are having os images and you are booting from any one of the disk at any given point of time<<<

"how can you say that " if you mirror copies...? you need either break the mirror " most of the case your "OS image disk will be internal" no "FC

(means ,second disk is unused as far as the os running on the first disk is concerned ) you can mount the lv's on the second disk<<<

if vg00" it mirror disk" how ???
everything possible "technically" if you do like that you may end up corrupting the data.

[johnsonpk]

The question is:
If I boot the server into disk1 OS, is there a way that I can mount the logical volumes sitting on disk2?

here is the steps in low level

assuming disk1 has hpux v3

and disk2 has v2

1) boot from disk 1
2)vgexport -v -p -s -m /tmp/vg00_v3.map vg00
cat /etc/fstab|grep vg00 >> /tmp/fstab_v3
copy /tmp/vg00_v3.map and /tmp/fstab_v3 to some other server as it wont be available after you boot from disk2
3)reboot the server and boot from disk 2
copy back vg map file and fstab to /tmp
4)create volume directory and group file
mkdir /dev/vg00_v3
mknod /dev/vg00_v3/group c 64 0x010000
5)import the v3's vg00
vgimport -s -v -m /tmp/vg00_v3 /dev/vg00_v3

6)create a mount point for v3 fs
mkdir /HPUX11iv3

7)edit /tmp/fstab_v3 and replace vg00 with vg00_v3 and prefix all mount point with /HPUX11iv3
8)append fstab with vg00_v3's information

cat /tmp/fstab_v3 >> /etc/fstab

9)mount fs
mount -a


now for eg /tmp on v3 vg00 will be available under /HPUX11iv3/tmp

repeat steps 1 to 9 for mounting v2 disks on v3 os

[Johnson Punniyalingam]

so many steps... dude..... I hope the end user level will be difficult.

>>3)reboot the server and boot from disk 2<<

again you see, why need take some many "dangerous" steps. again technicaly can be done as i said. It been always said . please d'not play / touch on your operating file system

so if were I would >>Newa<< I will not do unless its test server, we can all play with hit. imagine if you have "Database" .etc. something important installed ??? will you intend do the same thing ??

above steps I have done with "EMC Bcv copies" hence "OS image" / we were booting from "EMC" diks .

[johnsonpk]

johnson>>This realy stupid "how can you say that " if you mirror copies...? you need either break the mirror " most of the case your "OS image disk will be internal" no "FC

I didn't know that you have had such a strong Psychic power to conclude that the author's Vg00 is mirrored , I didn't say any thing about a mirrored root vg , his case is an exceptional and there are two disks and each disk is installed with two different version of hp ux,and i believe any one who are NOT "DUMB " reading his post will under stand that, He is not saying anything about mirrored root vg ! is he??

(means ,second disk is unused as far as the os running on the first disk is concerned ) you can mount the lv's on the second disk<<<

Are you nuts..." if vg00" it mirror disk" how ???

I am not discussing about if and else condition here as the author is very much clear on his post .

Also think twice before you use that kind of words , I know the correct reply to your comment but i wouldn't use here as I respect this forum as every one .Off course there will be some one exceptional cases :-)

[johnsonpk]

>>again you see, why need take some many "dangerous" steps. again technicaly can be done as i said. It been always said .

Lot of new ideas and broadcasts today dude..
I didn't know that a preview mode vgexport and vgimport ,mount are such a dangerous step.( may be for newbie who doesn't know much about LVM" ..

>>please d'not play / touch on your operating file system

I do agree here ..mounting those lvs and playing around those data may make that disk non-bootable/corrupted.. but Newa need it

[Johnson Punniyalingam]

>>( may be for newbie who doesn't know much about LVM" ..<<<

Yes >> Mr. Johnsonpk << You are the Genuiss in LVM. next time "if you want to reply" to Authors thread, d'not incude my line while you are posting, if you can advice the author. go directly "put information"

from my point of view, its my suggestion .. . Again .. (Its Author choice to pick my post if its wong or not) if he like it.

Not you.......lol

[johnsonpk]

Dear Johnson Punniyalingam,
let me clear my stand here

Author is asking a possible way for mounting lvs on one of the local disk/vg and your simple suggestion was to go for shared volume group
which need SLVM and configure cluster daemons .Even if you have all these set up ,how u are going to share a local VG?at first glimpse it does not seems to work in the current scenario ?Can you explain bit more ??

coming to my suggestion ,doing a vgimport with second disk, for me its seems to be simple and straight forward but you says it is wrong and dangerous..could u please explain why it is dangerous ? Also if possible ,please explain on what basis you are saying it is wrong


Also you have got every rights to prove me wrong if i give a wrong suggestion but with valid reasons and I believe everyone in this forum does have that :-)


I believe everyone here will have different logic and suggestion ,some might be right and some might not , Suggestion and argument between fellow members will improve our knowledge as well as the reader's but it should be on healthy ground


Regarding quoting your line and commenting on .... As this is a public forum and everyone here can express their views/suggestion and prove something is wrong or right with/with out proper supported details , you may consider opening a private forum to avoid commenting on your-own posts:-)


Appreciate your understanding ..


Johnson

[Roopesh Francis_1]

Hey guys Where is the confusion..
I totally agreed with Johnsonpk here. Here is my understanding
Disk1 has HP UX v3
Disk2 has HP UX v2

If you boot the system from disk1,server doesn't know the lvm layout in Disk2 until unless it is part of HPUX v3 /etc/lvmtab and once you import the vg in disk2 as stated by Johnsonpk ,it will update the /etc/lvmtab and server recognize the lvm information in Disk2 and all the Logical volumes will be visible.

The same steps you need to do if you boot from Disk2.I think there is no confusion about disk lay out as we assume both disk has LVM version1(if it different also there should not be any issue when you boot the system on HPUX v3).

Mr. Johnson Punniyalingam,

>---- "how can you say that " if you mirror copies...? you need either break the mirror >" most of the case your "OS image disk will be internal" no "FC

>(means ,second disk is unused as far as the os running on the first disk is concerned ) you can >mount the lv's on the second disk<<<

>----..." if vg00" it mirror disk" how ???

I don't think this is the forum where you can use these type of words. if you are not convinced with johnsonpk â s suggestion, why dont you try to justfy your comments rather than abuse his words

If you really don't know rules of conduct in this forum,below is the link.

http://forums11.itrc.hp.com/service/forums/helptips.do?admit=109447626+1266848169378+28353475

Cheers!!
Roopesh