Assistance needed in correcting jbod array pv, vg, lvol config

[Doug_3]

I posted an lvcreate error problem which was quickly resolved by various folks, but now I want to correct, if possible my original config.
I have a FC10 JBOD, I used mirror-ux and extent based stripes with strict and contigous parameters. I wanted to spread I/O across the 5 disks and 2 channels. Instead, pvdisplay -v shows the extents un-evenly spread. I have about 25 2 gig raw lvols belonging to informix dbspaces.
Questions:
1. Can I correct this at this late date and if so how? (I want to spread I/O across the physical disks and would prefer to get the stripe down to a decent size vs the 1 meg extent (I could use Informix mirroring if necessary). I am assuming I would need to break the mirror, alter the lvols, vg's and pv's, which sounds like an experiment in terror.
2. If not I could possibly re-load everything from the informix side once I had the jbods set up correctly. How would one of you guru's do this?

Thanks in advance,
Doug

[Bryan D. Quinn]

Hey Doug,

My personal thought and what I would do if I were in your situation, is I would completely redo the config. Meaning I would get a couple of really good backups and then start fresh in creating my structures. Then after they were created and I was happy, I would restore the data back to their respective lvols.

That would be my solution, just to make everything nice and clean. You do have risk in there with your backups/restore as usual.

Hope my two cents helps!
-Bryan

[Ross Zubritski]

I would tend to agree with Bryan, if you have the outage window and verified good backups, lay the sucker out right and lay it back down.

Regards,

RZ

[Michael Steele_2 · Accepted solution]

Sounds only possible when adding more disks via lvextend and increasing the stripe size, in your case 5 more disks.

http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0x43d299f08a51d61190040090279cd0f9,00.html