Data recovery emergency on a downed server... Help please!!

Madison Kelly linux-5ZoueyuiTZhBDgjK7y7TUQ at public.gmane.org
Tue Dec 30 21:09:37 UTC 2003


Hi again Byron,

   Oh, I would say I know a little about RAID... ;)

   I don't ship a server without -at least- RAID 1 and the new 
replacement server does indeed have two Seagate Barracuda 7200.7 SATA 
120GB drives in software RAID 1 but alas the server that dies was 
inherited and built on "budget" parts...

   I am also implementing a tape rotation on a new DDS4 drive where 
there is a tape for Mon-Thu then four Fri tapes to be rotated. Others 
have recommended the same scheme like you, and I have personally 
implemented this scheme for years and it works like a charm...

   I'll post an update in a sec...

Madison

Byron Sonne wrote:
>> Madison (who has passed the 24hours mark and wishes she could just get a
>> break and then go home!!)
> 
> 
> Please let us know how this all turns out. I'm quite interested in the 
> ultimate resolution! Sounds like a text book case of the typical failed 
> drive/bad tape nightmare scenario.
> 
> I've been through this kind of hell but it was when I was part of the 
> dark side and had to restore MS boxen, or attempt to, from bad 
> Veritas/BackupExec and/or Legato backups (though most of the time it 
> worked). Part of the overall procedure (which mind you wasn't followed 
> regularly for whatever reasons) was periodic restores to make sure 
> everything would work when called upon. Nothing like having a shelf full 
> of backups that turn out to be shitworthy in the end.
> 
> Oh; I would avoid CDs for backup... I've used numerous brands across 
> multiple writers and I'm finding that they are just not trustworthy 
> enough over time. Some people seem to have something against tape, but 
> I'd be far more inclined to trust DLT over CD. And keep them drives 
> clean! It does help.
> 
> My experience with using RAID (level 5 or similar levels of redundancy) 
> is that it obviates the need for 90% of the restores I would have had to 
> do in the past. Except, of course, people that want a restore done to 
> get an old version of a file back. RAID is not a panacea, but who gives 
> a shit about a busted drive or a 12 hour restore when you can just pop 
> in a new drive with nary a hiccup? If you got the money, go RAID and 
> complement it with a good backup system (I like DLT). Given a solid and 
> well designed IT infrastructure, using good stuff, backups should be 
> solid, reliable, and never, ever needed. Or for those with massive 
> amounts of data, go SAN.
> 
> While I'm rambling on, make sure you have clean electrical power, 
> preferably backed up with a UPS. Numerous restores I've done in the past 
> would have been avoided if the machine didn't hiccup or crash due to a 
> brownout or power failure. They don't have to keep your boxen running 
> for days on end, just enough time to close open files and back down 
> gracefully. They also help for restores and backups; rather annoying 
> when a long, long restore craps out midway through due to a power 
> glitch. Tapes will also be more reliable if your power is clean too, but 
> I suppose that is conjecture.


--
The Toronto Linux Users Group.      Meetings: http://tlug.ss.org
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://tlug.ss.org/subscribe.shtml





More information about the Legacy mailing list