[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: OS Reliability
Ate its filesystem, anyone else see a t shirt idea there??
Brian
----- Original Message -----
From: "Chris Palmer" <chris@nodewarrior.org>
To: "Bryan Irvine" <bryan.irvine@kingcountyjournal.com>; <misc@openbsd.org>
Sent: Monday, August 25, 2003 8:56 AM
Subject: Re: OS Reliability
> Bryan Irvine writes:
>
> > http://www.usenix.org/events/hotos03/reports.html
>
> Nice one.
>
> > This got me thinking, has anyone ever seen OpenBSD crash as a result
> > of something that can proved to not be a fault of the admin or
> > hardware? I haven't. In fact the only time I have ever "had to"
> > reboot was when cdrecord locked up the cd-rom and wouldn't eject (not
> > the fault of OBSD of course).
>
> Your locked CDROM problem is OpenBSD's fault, and probably also
> cdrecord's fault. Our expectation is that the computer will always
> respond to our commands.
>
> I have a similar problem with my digital camera: If I unplug the media
> reader before umounting the filesystem, OpenBSD will never shut down
> (actually, I never let it try for longer than 10 minutes; maybe someday
> it would have timed out and given up on that AWOL filesystem). Not even
> plugging the camera back in works. One of my forty-eleven software
> projects is to look into that someday...
>
> (Yes, I know I should always cleanly umount filesystems. I also know
> computers should adapt to people, not the other way around. Please flame
> me privately -- I have a feeling the list does not want to hear it.)
>
> I've never had any problem with OpenBSD failing randomly, though. The
> reason I am so maniacal in my support of OpenBSD is that, of all the
> many operating systems I have tried, OBSD gives me the least amount of
> trouble -- by far. Even if it only offered one feature, the fact that I
> need to apply only ONE security patch roughly every 12 months would be
> sufficient for me.
>
> Sometimes I think SMP would be nice to have in OpenBSD. Then I look at
> my Linux system, which ate its journalled filesystem and panicked
> because I had the temerity to run two CPUs and to use a particular
> kernel version that came out on Thursday instead of the normally safer
> Tuesday version.
>
> I blame Java.