Maybe I'm unique in this regard, but as an admin, if something goes down on one of my servers, I want it to stay down until I intervene.
Firstly, if I'm properly monitoring the process, then I'll be alerted and can investigate.
Secondly, there may a *reason* the process goes down, and having it down may be a good thing. If someone's trying to fuzz our httpd process for exploitation, then it happily restarting will open up a wider attack window.
Autopilots on production servers seem like a bad idea to me.
Right now there is no remote monitoring of services within systemd, but this can, for sure, be changed. One could write a daemon for systemd that monitors all the services that are started or should be started and send some notification to the admin's desktop system, possibly including the latest log output and other stuff. This could extend current monitoring systems. If a process locks but does not exit, systemd would not notice, but a "pinging" monitoring system would, because it checks availability of a service every one minute, for example.
Slashdot Mirror
Maybe I'm unique in this regard, but as an admin, if something goes down on one of my servers, I want it to stay down until I intervene.
Firstly, if I'm properly monitoring the process, then I'll be alerted and can investigate.
Secondly, there may a *reason* the process goes down, and having it down may be a good thing. If someone's trying to fuzz our httpd process for exploitation, then it happily restarting will open up a wider attack window.
Autopilots on production servers seem like a bad idea to me.
Right now there is no remote monitoring of services within systemd, but this can, for sure, be changed. One could write a daemon for systemd that monitors all the services that are started or should be started and send some notification to the admin's desktop system, possibly including the latest log output and other stuff. This could extend current monitoring systems. If a process locks but does not exit, systemd would not notice, but a "pinging" monitoring system would, because it checks availability of a service every one minute, for example.