Here is the orginal question,
Admins,
What is the safest way to replace a failed disk drive in an LSM mirrorset.
We are Using 4.0f,
the effected volume is part of an advfs domain.
The volume comprises two plexes each made by one sd
that is actually a stripeset on an HSZ 70 controller.
A volprint -Aht reports
one of the plex's as disabled and nodevice.
Thanks for your support in advance,
The majority of responses were like the following
>From Tyler Hall
voldiskadm is probably the easiest/safest bet.. it is a little interactive
menu cli.
Remove disk for replacement
add new disk back in
Tyler
This response was also very usefull
from Pat O'Brien
I always do a voldiskadm and select the remove disk fore replacement option,
fix or replace the hsz70 back to full strength, disklabel -z disk contianing
sd, then -wr sd hsz70, and then voldiskadm to re-add the disk for the
previously failed disk, and then the plexes go into a big syncronization
which takes about 3 hours to complete. I ask the users to minimize i/o
during this process to prevent delays. there is a voldiskadm trick, and
that the disk name to be removed is not the rz# but the lsm-disk# which may
be the rz# if that is the way you assigned them. voldiskadm will tell you
the rz# does not exits if you don't use that method.
This proved usefull, always remember to read carefully for what the menu driven
input asks for. I have tried several different methods, but this is easiest so
far....
Thanks again,
Lee Brewer
Received on Mon Sep 17 2001 - 14:27:55 NZST