On Thu, 26 Apr 2001, Zane H. Healy wrote:
> Hmmm, actually come to think of it, I've got a Linux Workstation at work
> now (finally had to give up the antique RS6000), and it's been up for
> nearly two months now. It's also running a modified RedHat 6.2 release.
The old Linuxhacker box was up for over 400 days running 2.2.10 or so
before the hardware started to get flaky (and the only reason it rebooted
before that was because of an extended power outage). I needed it to be
rock solid reliable because it was located on the other side of the
country to me. I had another machine here which ran 2.0.33 with never a
glitch for over a year before I decided to retire it in favour of
something faster. I personally don't recall ever experiencing any crashes
with stable kernel releases that weren't later traced to bad hardware,
(very occasionally) a buggy experimental driver for something, or (in the
case of ARM Linux a couple of years ago) a bug in GCC generating bad code.
The company I work for, an ISP, uses Linux for all it's servers and the
only problems they've experienced were with a buggy multiport 10/100
ethernet card driver which occasionally locked the card (the machine was
still usable but it needed to be rebooted to reset the card), and the
(experimental) kernel NFS server locking up (it seems to work find in
recent kernels though). Although I have often run development kernels on
my workstation, and I have done kernel development myself which sometimes
resulted in crashes, I've never seen it result in filesystem damage worse
than a couple of files which were being written to at the time of the
crash being left in /lost+found (only very rarely did it need a manual
fsck at all in fact, especially if I ran a sync before inserting the
potentially buggy kernel module that I was working on).
--
------- Alex Holden -------
http://www.linuxhacker.org/
http://www.robogeeks.org/
Received on Fri Apr 27 2001 - 04:43:10 BST