On Wed, Mar 05, 2014 at 10:25:28PM +0100, Martin Pala wrote: > > It's still racy. What if both systemd and monit decide to restart > > the process? Suppose, that systemd do this first, process was restarted > > and only then - monit request to restart arrive. Or, only systemd decide > > to restart the service, and while it doing this - monit coming with test > > to decide: bah, we should restart this service, it's dead... > > Yes, the situation which you described may happen - systemd should be able > to handle the case (abort the restart request if it's pending already or > queue the second restart request - which makes less sense).
I don't think it can, especially the second scenario. That's just an example that there are good reasons to make each program do one thing well. > In general you're right - it's not good practice to have two > independent pro-active process checks ... however as mentioned, Monit > can be configured to not act if process failed - it will still detect > that the process crashed, but won't do any action. Or just to work as a nice web gui :) That's why the monit package doesn't conflict with systemd (or upstart, whatever). As well, there are reasons why I suggest sysvinit. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
