On 07/01/2018 17:45, Melleus wrote: > Melleus <mell...@openmailbox.org> writes: > >> Neil Bothwick <n...@digimed.co.uk> writes: >> >>> On Sat, 06 Jan 2018 18:46:25 +0200, Melleus wrote: >>> >>>>> What do the logs say? >>>> That's all I could find in syslog: >>>> >>>> connmand[6709]: Aborting (signal 11) [/usr/sbin/connmand] >>>> >>>>> Can you start it manually? >>>> >>>> No, it pretends to start but fails silently. >>> >>> Looking at the man page, try adding --debug=DEBUG and --nodaemon >> >> Thank you for helping me. >> >> --debug=DEBUG is almost silent, but just --debug is more verbose. >> >> All I see is that something wrong is happening here: >> >> connmand[2434]: src/iptables.c:__connman_iptables_append() -t mangle -A >> connman-INPUT -j CONNMARK --restore-mark >> connmand[2434]: Aborting (signal 11) [connmand] > > Thanks again for pointing me to logs. Those iptables was a > problem. There are the closed bug #573174. Iptables higher than 1.6 > breaks connman. The solution is to use iptables lower than 1.6 or > connman higher than 1.32. So the combination of connman v1.29 and > iptables v1.6.1-r2 just cannot work. But unfortunately for me they both > have stable keyword. I beleive that this is a some kind of bug. > After I masked iptables higher than 1.6, reemerged the packages and > reboot, everything works like it should. > > I don't know whether developers are reading this thread, but it would be > very nice to keyword only v1.4.21-r1 of iptables with stable keyword > (like they have done with kernel recently) or promote to stable some > version of connman higher than 1.32 upstream. This would completely > have this bug eliminated even before someone other than me hits it.
Post your finding to b.g.o. It's a simple matter to limit which versions of iptables can be used with each version of connman. Tracking that, and making changes when they become known, is what being a package maintainer is all about. -- Alan McKinnon alan.mckin...@gmail.com
