On Mon, 4 May 2015 11:51:08 -0400
Matthew Miller <mattdm(a)fedoraproject.org> wrote:
Actually, *systemd* is that process. That's one of its huge
features
missing in the old init system. Take a look at the man page for
systemd.service — search for "Restart=".
However, we don't configure most things that way out of the box...
yet. There was discussion about this three years ago, reopened a year
ago...
https://fedorahosted.org/fpc/ticket/191
and the current guidelines
https://fedoraproject.org/wiki/Packaging:Systemd?rd=Packaging:Guidelines:...
but as far as I know there hasn't been a big effort to bring existing
services in line. (Seems like a worthwhile project if anyone's
interested.)
Those were interesting reads. It seems the infrastructure is in place
to deal with the problem of service failure, but operational inertia is
slowing adoption. In other words, just a normal human system. :-)
The one thing I didn't see addressed was the comment requesting some
notification if a service became a chronic (ab)user of the restart
feature, so remedial action could be taken or requested.
It should be a simple matter for the OP to put the recommended
Restart=on-failure command into the service file under the [Service]
section for the failing services and see what happens.