> We would like to know the recommended procedure to replace a failed and/or
> failing drive on 2 - 6800's with two multi-pathed A5000's. We have
> experienced multiple devices when doing a "vxdctl enable", thus not being
> able to import the drive back into Veritas control - Unless we reboot the
> servers with a"boot -r" (last resort). If the drive is dead and/or
> failing.....
> Does luxadm remove_device & insert device need to be run on both servers
> simultaneouly?
> Thanks.
Yes, if its to work properly, you should do it simultaneosly in that the
luxadm command will update paths and inof for a5000.
However, it can be done out of sync, but its just better to fix both at
once, normally no downtime is needed for this, i.e. everuthing is mirrored
or raid5.
Main question is, why do they share A5000?
If it is a cluster, procedures shoudl be written for this at site.
If not, I assume the two 6800 are mirroring their separate data, normally
I would have used the A5000 in split mode and made the systems just see
their disks, but perhaps this is not possible?
Otherwise...
1) disable disk from veritas use on both hosts.
2) start luxadm remove_device on both hosts
3) remove the disk
4) finish remove on both hosts
5) luxadm insert_device on both hosts
6) insert a new disk
7) finish insert on both hosts
8) vxdctl enable on both host
When I say on both hosts, it doesnt mean exaclty the same time, because
that will just confuse the A5000, in each step...
A) do host A, wait until it finishes,
B) do host B, wait until it finishes,
C) continue to next step.
Works for me.
If the disk is faulty in that it doesnt answer, you might have some
problems, but I wont go in to the resolution if thoose problems here.
/Johan A