[cobalt-users] RAQ2 Disaster Recovery

["John Burgess" <john.burgess@xxxxxxxxxx>]

Hi all,

I thought you might want to hear a tale of woe with a happy ending.

Had a production Raq2 go down last night due to failure of the Quantum
Fireball 12 gig HD.  It was barely 6 months old.

Luckily I had a spare Raq2.  Pulled out the failed HD, got a Y power
connector and a ide cable with two drive connectors, and put the failed HD
in a working system as a slave.  fsck'd the fail drive, and mounted it (read
only).  tarred the user mail and sites up to the working drive.  copied off
lots of other little files from /etc/sysconfig,
/usr/admserv/html/.cobalt/siteList, /usr/admserv/html/.cobalt/siteManage,
/usr/local/majordomo, /etc/sendmail.cw, /etc/virtusertable, /etc/passwd,
/etc/shadow, /etc/logrotate.d, /etc/aliases, /etc/aliases.majordomo,
/etc/httpd, /etc/hosts, some /usr/local things, /etc/mail/access, and there
were probably a few more but that is mostly it.

The server came back up, no major.  Everything working.  Ok.  So now I don't
want to wait forever to get my spare/sandbox back up so....

Went to Office Depot (they are close and cheap). Got a new Western Digital
12 gig hd, set it as a slave, took out the bad drive (that was setup as a
slave in the working box), and put it in the working server.  Tried a
straight dd from primary to secondary drive (hda to hdb), but that didn't
quite work (I don't know why).  So then I created partitions with sfdisk
(fdisk and cfdisk are no good for this) using sector specs as follows (these
come from a factory Cobalt drive):

   Device Boot    Start       End  #sectors  Id  System
/dev/hda1             1   1024127   1024127  83  Linux native
/dev/hda2       1024128   1287215    263088  82  Linux swap
/dev/hda3       1287216   1697471    410256  83  Linux native
/dev/hda4       1697472  25429823  23732352  83  Linux native

The I did the dd from partitions on hda to hdb (i.e. dd if=/dev/hda1
of=/dev/hdb1
bs=512) for partions 1, 2 and 4.

My work was rewarded with a working (and bootable replacement drive for the
Raq2).  Put it in the system that had the drive failure and deleted the
sites, and I once again have my spare/sandbox working as well as the
production system on-line.

I hope this helps someone else.  There is bound to be a lot of this disaster
recovery going on out there (especially with those Quantum HD's and those
itty bitty fans).

Now who do I send that failed drive to for replacement?  Should I even try
Cobalt or would Quantum give me quicker satisfaction?

John Burgess
President, Allegro Data Systems
12500 Network, Suite 401
San Antonio, TX  78249
210-558-0709
E-mail <mailto:john.burgess@xxxxxxxxxx>
Fastex.net <http://www.fastex.net/>