[collectd] collectd dies (sometimes)

[Sergiusz Pawlowicz]

On 5/9/07, Florian Forster <octo at verplant.org> wrote:
> Hi,
>
> On Wed, May 09, 2007 at 05:02:43PM +0100, Sergiusz Pawlowicz wrote:
> > collectd dies sometimes, once day, once a week, i used to manage
> > daemons by djb-daemontools, but i do not have them on every box, do
> > you think that i can put collectd in inittab, e.g.:
> >
> > co:345:respawn:/usr/sbin/collectd -f
>
> I'd be much better if you could enable writing of a coredump
> (`ulimit -c unlimited`) and, when the daemon dies again, send in the
> generated corefile, the binary (../sbin/collectd) and the plugins
> (../lib/collectd/*.so), preferably packet in some kind of archive, e. g.
> tar.
>
> With these files at hand we can determine _where_ the problem occured
> and hopefully fix it, removing the need to respawning the daemon.. ;)

sure, i can, but i need stable solution. so i can do both things, supervising
it and core analyse :-) so is that solution with inittab acceptable
for production
machine? what do you think?

s.