In article <Pine.NEB.4.53.0608022048250.373@agnus.datazap.net>,
DataZap  <az@datazap.net> wrote:
>Hi,
>
>I have a server that is still running 1.5.4 in a colo. I really need this
>system to run, because it is really hard for me to setup another newer
>server right now. Hopefully, I will have one soon (running NetBSD current,
>of course), but I am not sure what to do in the mean time. The server
>crashes every few days. Sometimes it will only have an uptime of 1 day,
>but other times it will have an uptime of almost 2 weeks. One thing that
>really gets me is that I run pine on this server and when I delete some
>emails as soon as I type "y" to confirm the delete it crashes. It has done
>this several times. I am not sure what system calls are done when pine
>deletes emails, but it seems to bring down the system. I thought maybe it
>could be because its is running out of file descriptors. So, I just built
>and installed a kernel with almost twices as many file descriptors. I also
>raised the limits on nmbclusters, maxfiles, and dfldsiz. I also built it
>with support for packet forwarding, because that is supposed to raise some
>kernel memory limits. I am also going to try an older version of pine
>(only because I can't get a newer version to build with the packaging
>system).
>
>The other thing that I am running on this system is thttpd. It seems use
>way too many file descriptors these days. I have the server restart thttpd
>every hour, because that seemed to help on my other box (my other box is
>running NetBSD 2.1), but on this box it seems to be going down more.
>
>Does anyone have any ideas on why the system goes down and what I can do
>to keep this from happening?
>

Does your kernel have ddb compiled in it? How does it crash? Can you
get a stack trace?

christos