[systemd-devel] Recommended way of running additional event handlers on coredumps?

[Lennart Poettering]

On Fr, 09.05.25 09:31, Johannes Barthel (johannes.barthel at farming-revolution.com) wrote:

> Hi,
>
> we're using an Ubuntu setup where systemd-coredump is set up as the coredump handler. This is fine, coredumps end up in /var/lib/systemd/coredump/. We would however like to additionally run our own event handler (for remote error reporting) in case of a process dumping core.
>
> Does systemd-coredump provide any facilities for registering an event handler for this? Or should we create our own handler, register that as kernel.core_pattern in sysctl and forward the coredumps to systemd-coredump? I considered subscribing to the journal and filtering the coredump event out there, but that might cause unnecessary CPU load and also the API seems to be kind of broken, I ran into this issue [1].
>
> What is the best way of running our custom error reporting script in
> addition to systemd-coredump's default behavior?

You could just define your own template unit, and then run that after
each systemd-coredump at .service instance. i.e. maybe name it
mysecondcoredumphandler at .service, and then do:

   mkdir -p /etc/systemd/system/systemd-coredump at .service.d
   cat >/etc/systemd/system/systemd-coredump at .service.d/mysecondcoredumphandler.conf <<EOF
   [Unit]
   Wants=mysecondcoredumphandler@%i.service
   Before=mysecondcoredumphandler@%i.service
   EOF
   systemctl daemon-reload

or something like that.

That means you service is enqueued whenever the primary handler is
invoked, and runs right after it.

Lennart

--
Lennart Poettering, Berlin