[Ian Jackson] > Even so, surely insserv shouldn't just spin. It should break the > cycle (perhaps badly) and be able to carry on.
It isn't spinning. It is called via update-rc.d from several postinst scripts, and report the problem every time. There is no way to know where to sensible break the cycle, so it was decided early to reject installation of packages introducing a cycle, to force people to address the issue right away while the machine is still running instead of risking a non-booting system to surprise the user on the next boot. The initial report just show a lot of packages being installed, not a loop. Sure, one could try to break the cycle, but the end result might be to brick the machine. Could this be caused by a removed but non-purged package with an broken init.d script? I've seen quite a few of those. -- Happy hacking Petter Reinholdtsen