LVM

[bagek]

I have a rp4440 HPUX 11.23. I am using a Smart Array 6400 controller. ciss4. We divided the ds2300 into 3 LD's. LD0, LD1 and LD2. these are RAID 0 mirrored to the identical LD's on another 6400 connected to the second channel on the ds2300 ciss6. LD0 reported a failed drive at channel 1 id 0 as FAILED. The lights on the hard drive looked ok. I brought down the system and replaced the suspect drive. It booted OK and SAUTIL reported the new drive as OK. However vgdisplay -v /dev/vg01 reports this warning "could'nt query physical volume /dev/dsk/c4t8d0" This is the id of the failed LD comprised of 2 36GB drives. Prior to this it was repoprted as UNAVAILABLE. ioscan reports c4t8d0 as claimed. Any ideas? Do I need to run some commands likie vgchange? thanks. 

[Torsten.]

Are the single disks configured as RAID0?

If so, you need to mark them as repaired (sautil) - then they are "CLAIMED" again.

[bagek]

yes. each LD is made up of 2 36GB hard drives which are then configured as a RAID0. This LD is then mirrored to 2 other 36GB disks which are also RAID0. diskinfo sees the new disk but pvdisplay -v /ev/rdsk/c4tod0 does not. Should I try vgcfrestore -n/dev/vg01/dev/rdsk/c4tod0.

[Redhat]

Yes as you replaced the diskk .You need to restore the LVM information by vgcfgrestore -n < raw disk> first

[bagek]

can i do this with the system up and running? will this cause any problems with our data? just being carefull. thanks. 

[Pete Randall]

> can i do this with the system up and running?

Isn't that the whole point of having mirrored disks?  Yes it should be safe.  Take a look at the "When Good Disks Go Bad" guide for your exact steps:

http://bizsupport.austin.hp.com/bc/docs/support/SupportManual/c01911837/c01911837.pdf

[Torsten.]

Since this is RAID0, the data is gone after a drive replacement.

The controller prevents access after this replacement.

You need to do this (example):

# sautil /dev/ciss3 accept _media _xchg 3

to make the logical drive accessibly again (here logical drive 3).

Then you can use vgcfgrestore, etc ...

[bagek]

OK. after the vgcfgrestore are therre any other commands to attach the disk to the mirror? thanks

[Pete Randall]

Sorry - I didn't catch the RAID 0.  I don't even understand why someone would use RAID 0.

[bagek]

this a weird system. I inherited it. Eventually I would like to swap the entire disk subsystem with a va array. What they did was to use a smart array 6400 ciss4 to take 2 disks and RAID0 them together to create LD0  and then use LVM to mirror this LD0 to another pair of disks RAID0 as LD0 on another 6400 ciss6 on the same ds2300 box. They split the ds2300 box into 2 channels. Does this help? sautil now sees the new disk as OK. I did do sautil /dev/ciss4 accept_media _xchg. Now I need to get it synced to the other mirror. What should I do after vgcfgrestore? Thanks

[Torsten.]

Do the vgcfgrestore followed by a vgchange -a y ... and a vgsync - this should help.

[bagek]

it looks like vgchange, vgcfgrestore and vgsync did the trick. thanks for all your help.