Jonathan de Boyne Pollard <j.deboynepollard-newsgro...@ntlworld.com> wrote: > Sven Hartge:
>> systemd happily runs "legacy" LSB init scripts >> > ... except when its one-size-fits-all approach does not work, of > course. Example: > * https://unix.stackexchange.com/questions/386846/ > This is the problem with even Mewburn rc scripts (as I can attest from > personal experience of writing replacements for an entire Mewburn rc > system) let alone with van Smoorenburg rc scripts (which are far messier > than Mewburn rc ones). One size does not fit all. One really is not > going to ever get a backwards-compatibility mechanism that copes with > all such scripts in the general case "happily". The problem here lies in most cases in the init-scripts, which where subtly broken to begin with. I myself had written such init-scripts, which worked well enough while used by SysV-init, but where technically not correct or did things in a way that worked on my system but would fail (or create a race condition) on other systems. After the switch to systemd those scripts failed on numerous ways and I cursed at first, but upon closer inspection I saw that they where broken to begin with and the faulty behavior could also happen with SysV-init. Fixing those scripts, or, where possible, converting them to native systemd.units was then the correct solution. Grüße, Sven. -- Sigmentation fault. Core dumped.