LVM problem

[Ramones]

Hi

I am confused! We've migrated 3 volume groups from one machine to another, that was some months ago...
Now I wanted to erase the vgs information from the original machine and I made a vgreduce instead of a vgexport. In the target machine the disks stayed unavailable on one of the 3 vg and the database stoped.
My questions are:
- Why the disks stayed unavailable only in one of the vgs?
- Can I recover from this with a vgcfgrestore? If so, should I do it on the target or on the original machine?
I hope someone had experienced a situation like this and can help me here.

Thanks in advance,
R

[TTr]

> We've migrated 3 volume groups from one machine to another, that was some months ago...
These VGs were not completely migrated if they were still connected to the original server. Leaving them still connected to the original server was a very bad thing. When you move disks from one server to another first you vgexport and then you disconnect them from one server and connect them to another.

>Why the disks stayed unavailable only in one of the vgs
Probably becasue the disks of the other VGs were not visible on the original server.

> Can I recover...
Yes. If you have a saved vgXX.conf file on the new server you should do it on the new server and forget about the original one.

[Court Campbell]

is there only one disk in each vg? did you run vgreduce on only one vg? If so that would be why the other vg's were ok. can you post a vggisplay -v from the target?

[Ramones]

Hi

We've managed to recover with vgcfgrestore. I know we should had disconnect the disks and made vgexport when we've migrated the vgs. The thing is we may need a roolback the day after...and we make alot of things in alot of machines at the same time...it was a mistake, it happens...the good news is that we could recover.
Thanks anyway...
R