disklabel differences FreeBSD, DragonFly

Mike Meyer mwm-keyword-freebsdhackers2.e313df at mired.org
Thu Jul 27 20:58:53 UTC 2006

In <20060727202105.GA14724 at titan.klemm.apsfilter.org>, Andreas Klemm <andreas at freebsd.org> typed:
> On Thu, Jul 27, 2006 at 02:28:18PM -0400, Mike Meyer wrote:
> > These days, the only technical reason I know of for having separate
> > mountpoints is because you want to run commands that work on
> > filesystems on the two parts with different arguments or under
> > different conditions.
> Well I still prefer to "design" my filesystems no matter how big
> disks are.

Yes, and most of those qualify as needing to "run commands that workon
filesystems on the two parts with different arguments ...".

> Also I'd gues that its still valid that less file movement
> in root filesystem increases robustness if you have a power outage.
> Same true for other important filesystems.

Actually, it one of the FS gurus convincing me that this was no longer
true that converted me to the idea that reducing the file system count
was a good idea. What you say was certainly true for 4.2BSD, but when
was the last time you saw an entire file system toasted when something
failed hard? Or even significant damage to files that weren't actually
in transit at the time?

I know the last time it happened to me - it was over a decade ago. I
had a desktop Solaris box I'd set up as web server because the IT
department was a typical IT department. So it wasn't on batteries like
every other server in the place. We took a power hit, and it crashed
hard. Power came back, and it started fsck'ing it's file
system. Partway through that we got a *second* power hit, so it
crashed hard in the middle of fsck'ing. This power outage lasted long
enough that the backup generators kicked in, so it came back up and
started fsck'ing again. At which piont the backup generators cut out,
because they hadn't been refueled after the last outage. So it crashed
hard *again* in the middle of fsck'ing the file system. The file
system was unrecoverable.

> Also it makes it easier to upgrade a system, since you only nail
> / and /usr, if the rest is in other filesystems.
> Also its easier to newfs "/" and "/usr", if "/var", "/usr/local"
> and "/usr/X11R6", "/home", ... are on differnet filesystems.

Right. I typically install / and /usr as distinct files systems for
just that reason (/ and /usr have different backup & recovery
strategies and I use dump, so that's why they are two partitions). So
why does / need to be different from /var, /usr different from
/usr/X11R6 and /home different from /usr/local? Seriously now - what I
just described is my typical install.

> You see, I think there is still demand for using many filesystems
> if you are open minded for having the best support in every "shitty"
> situation ;-)

Well, there are always special cases. But hardware is so cheap these
days, I'm used to fine-tuning the *system*, not just the partition.
If something is so critical that it needs it's own partition, it's
probably so critical that it needs it's own system as well. In fact,
most of the thing I work on these days are so critical that they need
several systems, half of them at a second site with automatic failover
between them.

Mike Meyer <mwm at mired.org>		http://www.mired.org/consulting.html
Independent Network/Unix/Perforce consultant, email for more information.

More information about the freebsd-hackers mailing list