[quagga-dev 4445] Re: zebra crashes do not dump core

James Aldridge jhma at mcvax.org
Wed Oct 18 10:42:15 BST 2006



--On 17 October, 2006 15:54:49 -0500 David Young <dyoung at pobox.com> wrote:
> It must have something to do with zebra reducing its privileges:
>
># ps aux -p `pgrep zebra`
> USER    PID %CPU %MEM VSZ  RSS TTY STAT STARTED    TIME COMMAND
> quagga 9511  0.0  1.7 488 1128 ?   Ss    6:47PM 0:13.35 /usr/sbin/zebra
> -d -f /
>
> (It's strange, but top(1) says zebra runs as root.)
>
># sysctl -a | grep core
> kern.defcorename = /var/core/%n.%t
> kern.coredump.setid.dump = 0
> kern.coredump.setid.path = /var/crash/%n.core
> kern.coredump.setid.owner = 0
> kern.coredump.setid.group = 0
> kern.coredump.setid.mode = 0600 (rw-------)
> proc.curproc.corename = /var/core/%n.%t
> proc.curproc.rlimit.coredumpsize.soft = unlimited
> proc.curproc.rlimit.coredumpsize.hard = unlimited
>
> According to sysctl(3),
>
>              kern.coredump.setid.dump
>                      If non-zero, set-id processes will dump core.
>
> If I change kern.coredump.setid.dump to 1, the kernel says that zebra
> *has* dumped core---it didn't say that, before:
>
> Aug  6 19:42:40 cuw /netbsd: pid 9511 (zebra), uid 1014: exited on signal
> 6 (core dumped)
>
> Nevertheless, there is no core in /var/core/.  I will have to explore
> some more.

Ahh... but kern.coredump.setid.path above is set to /var/crash/%n.core. 
Does the /var/crash directory exist?  If so, is there a 9511.core file 
there?

James







More information about the Quagga-dev mailing list