[svlug] new disk is installed - raid

Rick Moen rick at linuxmafia.com
Thu Sep 21 18:56:16 PDT 2006


Quoting Alvin Oga (alvin at mail.Linux-Consulting.com):

> > RAID1 is not -- not _ever_ -- a substitute for backup.
> 
> or raid1 is never to be used as "backup" ...
> not even considered for "substitute" .. "never" trumps substitute ?

I'm afraid I don't understand the question.

Redundant storage is simply _not_ backup.  Some people object to using
disk redundancy on grounds of it not being sufficient as a means of
backup:  I always tell them, you're right, but it's not _intended_ 
as any type of backup method.  (They have made a category error.)

The threat model against which disk redundancy protects is different.

I'm sorry, but is there any actual question of fact, here, or is this 
just you not liking the technology?


> it works as described if and when,
> people have raid properly configured and tested prior to initial deployment
> 	- "properly tested" is where most folks fail

I'm sorry, but what's your point?  I'm not understanding, since this
seems completely unresponsive to the quoted text from my message that
had preceded it.


> > And, by the way, I'll believe in SVLUG having reliable 4-hour backups
> > when I see it.
> 
> vi /etc/aliases
> 
> 	svlug:	svlug at backup.svlug.org
> 
> 	--->> all incoming emails is backed up within seconds
> 	though is a whacky way ... it doesn't lose any emails during
> 	the 4-hr time period of no backups

I sincerely hope you do not think that in any way resembles system
backup (and you have also completely ignored my point, in that we have
nothing even _that_ pathetic).

> >  You don't want to know about the group's backup
> > practices, trust me.  It'll frighten you.
> 
> if you mean there is no backup ... its not that frightening in that
> lots of people do not have backups

Is this intended as deadpan humour?  I hope it is.  Losing years of work
would be both frightening and depressing.


> > The worst that can happen with a RAID1 "md" pair is that you don't use
> > that capability, in which case all you lose is the purchase price of a
> > second hard drive.  Which in this case I'm offering.
> 
> nah .. worst case ... i've seen people do is reinstall from scratch
> because the single disk replacement thingie didn't work so they lost
> it ( on both disks ) and whatever else that was not backed up

Then, they fscked up, didn't they?

Can you explain to me why someone who's just lost half of a mirrored
pair would not immediately do a full system backup, as the first step?

If your answer is "Because they didn't know what they were doing", then
I submit that you are wasting our time with hypothetical personal
problems that have nothing to do with the technology under discussion.

Your hypotheticals involve people doing something that destroys the
remaining drive.  That is frankly pretty extremely inept with "md", 
but no more than a temporary embarrassment and annoyance as long as you
follow common sense and do a backup as the first step.

> 	- in which case, why bother with raid if it doesnt work

It did work.  They did something stupid that shot the other drive in the
head.  At that point, they revert to the backup they _obviously_ made
immediately before doing any operation that involves partition tables,
i.e., a RAID1 remirroring.

If they didn't do that backup, then _there_ is your problem.

This really is a poor use of our time, Alvin.





More information about the Svlug mailing list