[SUMMARY] Solstice DiskSuite question
- From: "J. A. Landamore" <jal@xxxxxxxxxxxx>
- Date: Tue, 28 Aug 2007 11:00:31 +0100
On 21st August I asked
--
I have 5 disks in a RAID 0 (stripe/concat) configuration under Solstice
DiskSuite. One is going bad (1 hard error, soft errors clocking up 2 or 3
a day) As you can see the system isn't mirrored, so if I'm not careful
I'll have to do a full restore of all the data, tedious.
The system is under contract, so a replacement disk is arriving tomorrow.
If I format the new disk the same as the old one, don't newfs it, and dd
the data partition from failing to new, what is the chance of it working?
--
Thanks to
Andrew Toon
James W. Abendschan
Darren Dunham
John Stoffel
Jeff Woolsey
Brad Morrison
for their replies and messages of support, especially those who mentioned
about SLVM keeping track of the disk ids.
In short it worked.
The technique is as follows
remove any state database replicas on the failing disk using metadb -d
comment out S35svm.init in /etc/rcS.d to stop volume manager initialising
comment out entry in /etc/vfstab for the metadrive
touch /reconfigure to force a configuration reboot
halt the system
connect the new disk
boot single user
format and label the new disk identically to the failing one. (Here I was
probably very fortunate as the disk supplied was same make and model as
the failing one)
dd the data partition from the old to new disk. I used a block size of
248b i.e. one track per transfer This stage must be error free.
touch /reconfigure
halt the system
replace the failing drive with the new one, making sure the SCSI address
stays the same i.e. if the failing drive was c1t2d0 the new one must be
c1t2d0
boot single user
update the disk id using metadevadm
uncomment S35svm.init in /etc/rcS.d
halt the system
boot single user
fsck the metadrive
If this goes cleanly then uncomment the metadrive in vfstab and reboot.
You can add state database replicas if you wish
This took me 90 minutes to rebuild the system as opposed to approximately
18 hours if I'd had to restore all the data from backup.
--
John Landamore
School of Mathematics & Computer Science
University of Leicester
University Road, LEICESTER, LE1 7RH
J.Landamore@xxxxxxxxxxxx
Phone: +44 (0)116 2523410 Fax: +44 (0)116 2523604
_______________________________________________
sunmanagers mailing list
sunmanagers@xxxxxxxxxxxxxxx
http://www.sunmanagers.org/mailman/listinfo/sunmanagers
- Prev by Date: fun with route
- Next by Date: Summary: Jumpstart Solaris 9
- Previous by thread: fun with route
- Next by thread: Summary: Jumpstart Solaris 9
- Index(es):
Relevant Pages
- Re: Help Failing Disk Problem
... }>>> and the FreeBSD OS on a hard disk that is failing. ...
What is the best way in FreeBSD to do that? ... }>> of rsync, which is the
nectar of the gods, but a lot of folks seem to ... (freebsd-questions) - Re: XP Pro and software mirroring
... the SATA HD is failing on my HP DeskPro dc7600. ... nearly every HDD manufacturer
provides either in the retail box of the HDD ... you have a disk imaging program such as
Acronis True Image or Symantec's ... Ghost you can use that type of software to undertake the
disk cloning ... (microsoft.public.windowsxp.general) - SUMMARY ADDANDUM: LSM recovering after error OK, what about FAILING in volprint ?
... LSM Commands Fail When Disk Left in Failing State ... pl cluster_rootvol-01
cluster_rootvol ENABLED ACTIVE 2995888 CONCAT - RW ... dsk24b ENA ... (Tru64-UNIX-Managers) - Re: Help Failing Disk Problem
... and the FreeBSD OS on a hard disk that is failing. ... What is the best
way in FreeBSD to do that? ... dump has the problem that a lot of tools have, though,
including rsync. ... (freebsd-questions) - Re: How to clone a failed hard drive
... If your hard drive is failing due to logical errors, cloning only copies the
... I have cloned several disks very successfully with Ghost. ... files and folders
and do a clean install on the new disk. ... (microsoft.public.windowsxp.perform_maintain)