Import volume group with missing PV's

[Paddy O'Connell]

I'm looking for help here guys, a quick explain.

I'm testing our failover, so what our standard config is as folows.
We add a 3rd SAN volume to the internal 2 PV's in VG00.
vg00/lvol1-4 are extended to 3rd SAN disk
We create a vg01 where we move vg00/lvol5-8 "out onto the SAN"

So my problem is when we failover i need to bring up the 3rd mirror disk in a volume group on its own. I need help doing this.
I'm using a EMC DMX for storage and i've split off the BCV of the 3rd mirror, so to the failover node it's RW. I can see it and device files are created.
I've run the vgchdid on the PV's but get the below
vgchgid: Invalid LVMREC on Physical Volume

I do the mkdir, mknod bit anyway and then i import the LUN but get the below

Beginning the import process on Volume Group "/dev/vg00locutus".
vgimport: Warning: Volume Group belongs to different CPU ID.
Can not determine if Volume Group is in use on another system. Continuing.
vgimport: Quorum not present, or some physical volume(s) are missing.

I've tried activating the device group with the vgchange -a y -q n /dev/vg00locutus.

I've moved the lvmtab and created a new one with vgscan

Anything you can think of to get this online.

Thanks
Paddy

[Court Campbell]

I really do not understand this configuration. There is no reason to extend vg00 onto a SAN in my opinion. and I really don't understand moving the lvols 5-8 onto the san. If you need two systems that look identical it would make more sense to create an ignite image of one of the boxes using the Interactive option. Then you could ignite the second box and make changes to the hostname, ip, etc., during the ignite process. Once done then you would install patches, etc on the boxes at the same time to keep them pretty much the same.

You would want to failover the data that needs to be accessible by both boxes in order to keep whatever service it is your providing up and running. An explanation of what it is you are trying to accomplish may help.

[Sandy Chen]

Hi,

I'm not really understand your configuration. But your real environment is a vg00 with 8 lvol and your mirror is seperated with a disk of 4 lvol and another with the rest.

I don't see there is a way to bring the disk up. Your original lvm config that is written in the first mirror disk if for 8 lvol meanwhile it can only see 4. i don't see how you can bring it up.

Regards,
Sandy

[Todd McDaniel_1]

I too am baffled by this configuration.

However, I see an issue with the vg00locutus name not having a slash after the vg00 part.

If vol5-8 are required HP-OS volume groups, then you cannot configure to this way.


I would be curious to know the reasoning for trying to divide your OS up in this manner.

There is no reason I can see to "vailover the OS". It sbould independently loaded on each host.

[Marcel Burggraeve]

Maybe it would help if you try to explain a bit more what you're trying to achieve and what you've done so far exactly ?

You say you added a 3rd san volume to the interal pv's in vg00.
Did you do this with a vgextend ?
lvol1-lvol4 are extended to that san disk, how did you do that ?
Why are lvol5-lvol8 on a separate vg on the san ?

[Torsten.]

It is a bit confusing.

"vg00/lvol1-4 are extended to 3rd SAN disk"

extended means mirrored or not?

"We create a vg01 where we move vg00/lvol5-8 "out onto the SAN" "

These lvols are now part of vg01?

[Andrew Young_2]

Hi Paddy.

I think it would be best for you to start again by restating what it is you are trying to achieve. Are you trying to have a server that is currently booting off internal disk being able to be booted off SAN as well? Or by failover do you mean that you will boot another server using the SAN vg00 of the first server?


Regards

Andrew Y

[Paddy O'Connell]

Morning Guys,

I'll explain the bizzare config:-), with a view to helping me bring this online.

This is completely inhereited so don't shoot the messanger

We've 2 data centres 5 miles apart with EMC and SRDF looking after the data between them.

We buy a production node for 1 site and it's test node in the other site. If we need to failover the test node becomes production.

Meaning we rename & ip address the test node to the same as production. failover the device groups which contains app i.e. the oracle databases and then we have a sys device group which is the vg01 volume containing the lvol5-8 so when we edit the fstab to but the test node as production it all comes up as production.

However, there is info we still need to get from lvol3 in particular, stuff like /tcb, /etc and so on.

Hope this sheds light on it

[Court Campbell]

My suggestion would be to invest in serviceguard.

[Paddy O'Connell]

Court all i can say is amen to that:-)