From: Urie, Todd (TUrie@trueposition.com)
Date: Tue May 28 2002 - 13:23:26 EDT
Thanks to the following people:
Darren Moulding
Dave, bela@nivek.org
Tom Payerle
Pierre Zimmermann
The following suggestions were offered:
#---------------------------------------------------------------------------
The metaoffline command effectively fails the disk slice in question. I am
not sure how convincing a test that is, of course.
Removing even hot swappable disks from a system before the system denotes
the
drives as "failed" can cause hardware damage, or so I have been told.
Having
done it a few times before learning of this, I believe it is one of those
one
time in a hundred or more type things, but worth bearing in mind.
When I wished to test my disksuite mirroring, I offlined all partitions then
pulled the disk, which was a reasonably good test in my opinion. Not
offlining
the disks would be even better, but chances are the OS could tell a disk
failed
(and presumably would automatically offline it), and the (albeit small) risk
of HW damage was enough to drop that idea.
As for SNMP traps, I believe this is done in conjuntion with another
package.
I just use a cron job running metastat and looking for any suboptimal
metadevs.
Tom Payerle
#---------------------------------------------------------------------------
Hi Todd,
try cfgadm_scsi or cfgadm_pci to remove a disk from your system (see man
pages of cfgadm)
e.g.: cfgadm -x replace_drive c0::dsk/c0t0d0
or cfgadm -cf disconnect c0::dsk/c0t0d0
be sure that you have also created a boot record on the second disk, if the
first one is your boot disk!
e.g.: installboot /usr/platform/`uname -i`/lib/fs/ufs/bootblk
/dev/rdsk/c1t0d0s0
and enter this as an alternative boot disk (second boot device) in the boot
PROM)
for logging we use syslog, disksuite writes its messages to syslog.
cu
Pierre
PS: there are already some usefull scripts:
http://pixie.madstop.com/jumpstart/scripts/install-disksuite.fin
#---------------------------------------------------------------------------
> When I have two disks mirrored, is there anyway that I can simulate a
> failure of one of them, short of just pulling the cable on one of the
disks?
Nuke the VTOC on the mirrored disk ;)
> Also, I thought that I read somewhere that DiskSuite will send SNMP traps
on
> failure, however, I can't seem to find where I read that. Is this true?
If
> so, can someone point me to the docs?
yep. Disksuite can send snmp traps when shit happens. The mib is kinda
limited in scope but it'll do the job.
http://docs.sun.com:80/ab2/coll.260.2/DISKSUITEUG/@Ab2PageView/idmatch(SERVI
CETASKS-30812)?DwebQuery=disksuite+OR+snmp&oqt=disksuite+snmp&Ab2Lang=C&Ab2E
nc=iso-8859-1#SERVICETASKS-30812
WOOO! UGLY URL!
-- Dave. bela@nivek.org #--------------------------------------------------------------------------- Todd, A little slow with this reply, but you never know you may not have got this solution yet. You could fmthard -s /dev/null /dev/rdsk/cxtxdxs2 (this should blat the contents of the disk in question, failing the mirror) Then use metareplace -e dxx /dev/rdsk/cxtxdxs2 (your original mirror and disk) Regards Darren Moulding PS I recommend you test on non live data 1st Todd Urie Unix Support Specialist 610-680-2231 Direct turie@trueposition.com _______________________________________________ sunmanagers mailing list sunmanagers@sunmanagers.org http://www.sunmanagers.org/mailman/listinfo/sunmanagers
This archive was generated by hypermail 2.1.7 : Wed Apr 09 2008 - 23:24:23 EDT