On 2015-05-06 12:29, Michael Biebl wrote: > reassign: -1 aiccu > > Am 06.05.2015 um 04:43 schrieb Daniel Albers: >> This is a critical bug of either aiccu or systemd as explained by >> Pascal. >> >> Assigning also to systemd as the aiccu maintainer considers this a >> systemd bug, although I think message #26 < >> 546673f0.1040...@localhost.localdomain.org> already points to the >> proper fix for aiccu. > > Jeroen Massar is not listed as maintainer of the aiccu package
Apologies for not being able to get a DD-bit. I am only the person that designed and implemented aiccu, for the rest indeed, that is totally a useless thing in the Debian universe, why care about the original author of the code. > and I > don't see any explanation which would hint at a bug in systemd. Adding systemd to it all started causing problems. But that is indeed likely just a symptom. systemd is all magic as this bug shows. As you point out though the 'ifup.d' script might be the cause of the whole problem though. See also bug #689584 where I've noted that that script should not exist. > Therefor re-assigning back to aiccu. I think it should be re-assigned to Debian then as the problem lies there as that is where the modifications where made and well, Debian is the entity that can really resolve this. > From a quick look at the package, it seems to install a init script > which depends on $network and an if-up.d hook which restarts the aiccu > init script as part of /etc/init.d/networking. > /etc/init.d/networking itself provides $network, so there is a dead > lock, since systemd evaluates dependencies at runtime. Should such a "dead lock" at least not be clearly noted in a log or similar? > Reinier, if you need help with getting this sorted out, please contact > the pkg-systemd team. He does not want to maintain the package: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=692465 Thus all those problems will remain. And without a DD-bit, nobody can resolve this problem. > Is there a reason, aiccu needs to be restarted on ifup? Absolutely not. There is no need to *EVER* restart AICCU. The protocols are made for handling dynamic networks. Unfortunately Debian package "maintainers" seem to think they know better hence instead of listening to the original author of the code and the persons who run the SixXS service (hint: same people :) things are changed anyway. Nothing we can do about these problems. That is a Debian issue. See also amongst others: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=689584 https://bugs.launchpad.net/ubuntu/+source/aiccu/+bug/223825 Note that Debian is not the only entity who does not get this, OpenWRT is another such entity, and heck even ZyXEL still does not understand the concept. > Does it need that to pick up new network interfaces? No. The AYIYA protocol is made for handling network changes. > The better fix is, to make aiccu network hotplug aware (e.g. via > rtnetlink) [1]. Why would AICCU need to know about interfaces? > This would also mean, aiccu would work better ootb with > other networking tools or if the network is configured manually. It already works fine in those situations. Greets, Jeroen -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
