Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 7 Oct 1998 16:38:39 -0700 (PDT)
From:      asami@FreeBSD.ORG (Satoshi Asami)
To:        mike@smith.net.au
Cc:        current@FreeBSD.ORG
Subject:   Re: CDROM as system disk
Message-ID:  <199810072338.QAA11608@silvia.hip.berkeley.edu>
In-Reply-To: <199810071347.GAA03046@dingo.cdrom.com> (message from Mike Smith on Wed, 07 Oct 1998 06:47:11 -0700)

next in thread | previous in thread | raw e-mail | index | archive | help
Guys, thanks for the pointers to various projects and products.  I've
saved them all and will post it here if I find something interesting.

 * You want to discover that CDROM drives have a similar MTBF under 
 * constant use?  They'd slower, have lower data density, and if anything, 
 * will fail more rapidly.

Eek.

 * I've played with it off and on.  You really don't want to do it for a 
 * production environment unless you never touch the system disk in normal 
 * operation.

Well, the system disk doesn't hold much more than just basic system
utilities.  In our case, /, /usr and /usr/local which contains apache
and gridpix servers.

 * The basic issue is that you still need some writable storage, and as 
 * long as that's local, you still have local disks.

We do have a lot of local disks.  We just don't want to boot from SCSI 
disks. :)

 * > (Incidentally, nologin seems to belong to /var/run in all senses of the 
 * >  word; does anyone know why it's in /etc at all?)
 * 
 * Hysterical raisins.

I take it that nobody will object if I move it to /var/run?

 * You have a reliability problem that you need to address.  To start 
 * with, you need to work out whether the problem is actually inherent in 
 * the disks you're using, or whether it's environmental.

I believe it's environmental.  Most of the failures we've seen are to
internal disks, not necessarily of the same brand (even some SCSI
disks).  The machine room is kept cool year-round but there isn't much
air flow inside PCs I guess.

 * Then you need to follow through; if its environmental (eg. heat,
 * humidity, etc.), fix it.  If it really is the disks, consider using
 * better disks.  Plenty of systems up 24/7 have disks 5 years old or more.
 * Consider using removable disk sleds and keeping some spares around.

It is kind of hard to fix without getting new PC cases.  I'd like our
solution to be generic as possible too.

 * Dumping disks and going with CDROMs isn't going to help you
 * reliability-wise at all, and it'll hurt you in lots of other ways.

It seems the knocks on CD-ROM drives are that they are not meant for
heavy use and could fail if subjected to continuous start/stops.
Maybe it will work better if I go with an MFS root and copy regularly
used stuff onto it, making the CD-ROM used only for rarely-used
commands.  Actually, given that the working set of binaries is very
small, wouldn't they be in the cache anyway?

Satoshi

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199810072338.QAA11608>