We've historically had several bugs that result in a service dying and becoming unavailable. We can therefore assume more such bugs will pop up in the future. This is problematic as ThinLinc can often be a critical service, and if it is broken then users might not be able to do their work. Right now, a sysadmin has to notice that the service is down and manually start it again. HA and load balancing provide some protection, but not all installations have that. And the bug might be something that affects too many nodes. It would be useful if the services could automatically restart themselves, in the hope that the bug was temporary, and hence retain functionality.
An easy way to solve this is to use systemd's restart feature. That also has added benefits, like automatic handling of giving up if the service keeps crashing.
One blocker of this is how to deal with the risk of the config having changed on disk, but not loaded in to the service. It could be very surprising if the service suddenly changes settings without any user involvement.
Note that it is now much less likely that the services die on bugs as of bug 7636. asyncio tends to catch and log most things, unlike asyncore.
