[systemd-devel] Restarting sshd

[Lennart Poettering]

On Tue, 11.01.11 09:39, Gustavo Sverzut Barbieri (barbieri at profusion.mobi) wrote:

> 
> On Tue, Jan 11, 2011 at 7:12 AM, Alexander E. Patrakov
> <patrakov at gmail.com> wrote:
> ...
> > ExecReload=/bin/kill -HUP $MAINPID
> 
> This becoming a common case, shouldn't we have a ExecReloadSignal=HUP,
> or SignalReload=HUP?

Gmrbl. I don't want to bless sending SHIGUP really as a good solution by
introducing ExecREloadSignal or something like this. The reason is that
it is asynchronous by nature, and there's no way to wait for the reload
to happen and finish. The start and stop operations for services are
always synchronous and so should the reload operation be. While in many
cases the distinction between async and sync reloading doesn't matter
much I definitely want to make sure that people think about this and
that they do things synchronously if possible and only fall back to
async reloading via SIGHUP or suchlike if it doesn't matter in the
specific case or there is really no other way.

Or in short: I'd much prefer if people would use dbus-send (or some
service-specific equivalent) to synchronously fire off a reload method
call instead of using an async SIGHUP signal.

Lennart

-- 
Lennart Poettering - Red Hat, Inc.