Re: systemd and its journey towards taking over the world !!!
"I never understood what 'problem' was being solved by systemd !!!???"
More concisely Poettering &co dosn't seemed to have understood the basic problems with systemv init.
Having constructed countless start script over four decades I found that you expend a lot of effort on basic boiler plate code:
- checking the service's binaries and configuration files are present
- checking the service isn't already running
- checking any prerequisites are installed, and if services, that they are running
- after starting the service taking care that the process is demonised, checking that is actually running
Even at that point if the service has to be monitored and restarted attempted on failure you have run another service.
A minimal systemd by itself did more or less address these issues. I found constructing unit files under RHEL7 a lot less labour and not unpleasant. Once systemd took over autofs (automount) the lunacy seems to have set in.
Alternatives for looking after the system start up and services weren't new. Before systemd - DJB's daemontools had been around since the Ark; Solaris' service management facility (smf) dates from 2006. Linux has had quite a few init replacement contenders before systemd and still has.
My main objection to the systemd circus is that it is not easily decomposable - if take one bit you mostly have to take the lot; including, unfortunately, the clowns.