Subject: Re: raidframe/raid reliability
To: Dan J Fraser <dfraser@capybara.org>
From: Herb Peyerl <hpeyerl@beer.org>
List: port-i386
Date: 04/13/2000 06:16:15
Dan J Fraser <dfraser@capybara.org>  wrote:
 > I need to colocate a machine and have it be somewhat reliable.  So, I was
 > thinking I should put two drives in it, and use RAIDframe with level 1.
 > 
 > The reliability level I want is this:  If a drive dies, the machine should
 > keep working until I can get on-site and fix it.  It doesn't have to be
 > hotswap, etc... A few minutes of downtime for a driveswap/reboot is fine,
 > but a few hours of downtime while I drive to the facility is not fine.

I use raidframe with level 5 on a machine in stockholm and it does that
for me.  I haven't played with '1'.  I don't even shut the machine down 
when a drive fails, but then I'm a little more daring.  Replace the disk
with another identical one and then run 'scsictl scan'.

of course, you have to watch the msgbuf to see if a drive has gone bad.