Re: systemd services that are not equivalent to LSB init scripts
Hi,
On Mon, Jul 15, 2019 at 01:49:04PM +0200, Guillem Jover wrote:
> > In the same way, we could implement "service monitoring" in sysvinit by
> > adding an "inittab.d" directory, but I'm fairly sure that I'm not the first
> > person who had this idea in the last thirty years, so there is probably a
> > reason why it hasn't been done.
> Yeah, this is something that has slightly bothered me too, even though
> sysvinit is a bit poor at services monitoring TBH. I guess I might either
> file a request upstream, or send a patch at some point.
That's why I put it in "scare quotes". My general thoughts on this are:
- service monitoring is itself a service that may have an arbitrarily
complex technology stack and dependency tree
- except for very specific circumstances, restarting a failed service is
the wrong thing to do
- files might be in an inconsistent or invalid state
- if the service was attacked, restarting it gives more attack surface
So my guess is that there was simply no real demand for init to ever
restart daemons in a real world scenario.
I've never needed it in twenty years, at least.
Simon
Reply to: