Re: [OpenWrt-Devel] dhcpv6.script being triggered a lot
On 02.10.2013 07:35, Nathan Hintz wrote: Hi Steven: netifd eventually crashed: Tue Oct 1 21:57:58 2013 daemon.notice netifd: wan6 (1382): Segmentation fault Could this be due to continuing to call system_add_address(dev, a_new); without ever calling system_del_address(dev, a_old);? Hmm strange, I tested this fix and fired multiple rounds of RAs at it but couldn't reproduce this segfault. Also calling system_add_address just sends a netlink control-message to the kernel and does not allocate anything which must be deallocated later (same goes for _handle_subnet_route) so this shouldn't be causing this. -Steven ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
Re: [OpenWrt-Devel] dhcpv6.script being triggered a lot
On Wed, 02 Oct 2013 08:21:33 +0200 Steven Barth cy...@openwrt.org wrote: On 02.10.2013 07:35, Nathan Hintz wrote: Hi Steven: netifd eventually crashed: Tue Oct 1 21:57:58 2013 daemon.notice netifd: wan6 (1382): Segmentation fault Could this be due to continuing to call system_add_address(dev, a_new); without ever calling system_del_address(dev, a_old);? Hmm strange, I tested this fix and fired multiple rounds of RAs at it but couldn't reproduce this segfault. Also calling system_add_address just sends a netlink control-message to the kernel and does not allocate anything which must be deallocated later (same goes for _handle_subnet_route) so this shouldn't be causing this. -Steven I searched for a core file, but didn't find one. Any suggestions for how I can help debug this if it happens again? It may be completely unrelated to your changes; although I haven't seen any crashes in 'netifd' in quite some time. I've reverted to the previous 'netifd' for now, but I will try it again tomorrow evening when I get home from work to see if it crashes again. -- Nathan ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
Re: [OpenWrt-Devel] dhcpv6.script being triggered a lot
Hi Nathan, actually it's not that easy. Especially if the contents are the same this means an update. This is because most contents of the RA contain a validity timer in seconds. Thus if you sent 2 successive RAs with the same contents with a 3 second difference the second RA increases the validity of the contents by 3 seconds (as it arrived 3 seconds later). Cheers, Steven ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
Re: [OpenWrt-Devel] dhcpv6.script being triggered a lot
On Tue, 01 Oct 2013 09:51:23 +0200 Steven Barth cy...@openwrt.org wrote: Hi Nathan, actually it's not that easy. Especially if the contents are the same this means an update. This is because most contents of the RA contain a validity timer in seconds. Thus if you sent 2 successive RAs with the same contents with a 3 second difference the second RA increases the validity of the contents by 3 seconds (as it arrived 3 seconds later). Cheers, Steven ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel True, but does it need to be passed on to netifd? Depends on who maintains/monitors the validity timers I guess. -- Nathan ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
Re: [OpenWrt-Devel] dhcpv6.script being triggered a lot
On Tue, 1 Oct 2013 01:30:07 -0700 Nathan Hintz nlhi...@hotmail.com wrote: On Tue, 01 Oct 2013 09:51:23 +0200 Steven Barth cy...@openwrt.org wrote: Hi Nathan, actually it's not that easy. Especially if the contents are the same this means an update. This is because most contents of the RA contain a validity timer in seconds. Thus if you sent 2 successive RAs with the same contents with a 3 second difference the second RA increases the validity of the contents by 3 seconds (as it arrived 3 seconds later). Cheers, Steven ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel True, but does it need to be passed on to netifd? Depends on who maintains/monitors the validity timers I guess. Does the interface need to be torn down and rebuilt just for a validity timer update (barring the case when the update makes it go invalid)? -- Nathan ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
Re: [OpenWrt-Devel] dhcpv6.script being triggered a lot
Hi Nathan, I just pushed an update to netifd. Could you please check if the current version improves behaviour for you? Also alternatively could you please check in your /etc/config/6relayd if compat_ula is set to 1 and if so remove it or set it to 0? Cheers, Steven ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
Re: [OpenWrt-Devel] dhcpv6.script being triggered a lot
Hi Steven: Thank you for taking the time to look at this. compat_ula was set to 1 in /etc/config/6relayd; setting it to 0 eliminated the logging every 3 seconds by the avahi-daemon in my down stream linux box. The change to netifd didn't have any effect on this, which I believe is as expected. With the updated netifd, similar logging by the avahi-daemon running on the router has been eliminated as well (it wasn't occurring nearly as often though); so the change appears to be having an effect. I am still seeing logging by dnsmasq on the router that it is re-loading it's configuration files anywhere between 0 and roughly 6 times per minute (dnsmasq isn't being restarted, as it's the same PID each time). The same dnsmasq behavior was occurring with the previous version of netifd as well. I will let this run overnight, and let you know if I see anything else. -- Nathan ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
Re: [OpenWrt-Devel] dhcpv6.script being triggered a lot
On Tue, 01 Oct 2013 19:42:35 +0200 Steven Barth cy...@openwrt.org wrote: Hi Nathan, I just pushed an update to netifd. Could you please check if the current version improves behaviour for you? Also alternatively could you please check in your /etc/config/6relayd if compat_ula is set to 1 and if so remove it or set it to 0? Cheers, Steven ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel Hi Steven: netifd eventually crashed: Tue Oct 1 21:57:58 2013 daemon.notice netifd: wan6 (1382): Segmentation fault Could this be due to continuing to call system_add_address(dev, a_new); without ever calling system_del_address(dev, a_old);? -- Nathan ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
Re: [OpenWrt-Devel] dhcpv6.script being triggered a lot
Hi Nathan, I've seen this being reported by someone else already a few months ago also involving Comcast ( https://forum.openwrt.org/viewtopic.php?pid=191916#p191916 ). It seems they have a rather weird way of sending out Router Advertisements every 3 seconds at some locations which is causing this behaviour. While 3 seconds is technically the lower bound of the interval range it doesn't make much sense to spam these control messages around at that interval so it might be a good idea to contact Comcast about it. Cheers, Steven ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
Re: [OpenWrt-Devel] dhcpv6.script being triggered a lot
On Mon, 30 Sep 2013 08:18:58 +0200 Steven Barth cy...@openwrt.org wrote: Hi Nathan, I've seen this being reported by someone else already a few months ago also involving Comcast ( https://forum.openwrt.org/viewtopic.php?pid=191916#p191916 ). It seems they have a rather weird way of sending out Router Advertisements every 3 seconds at some locations which is causing this behaviour. While 3 seconds is technically the lower bound of the interval range it doesn't make much sense to spam these control messages around at that interval so it might be a good idea to contact Comcast about it. Cheers, Steven ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel Thanks Steven! I've posted the question on the Comcast support forums. If the content of the RA is completely unchanged, should odhcpc6 be treating it as a duplicate and filter it out? -- Nathan ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
[OpenWrt-Devel] dhcpv6.script being triggered a lot
My cable provider (Comcast) recently enabled native IPV6, and I've enabled it on my router (using a recent trunk version). While generally the IPV6 connectivity is working, I am seeing something strange. Roughly every 3 seconds, the dhcpv6.script is being triggered to run (by odhcp6c). My directly attached linux box (running Fedora 19) eventually fills it's logs with messages from avahi-daemon with the following msg (at roughly 3 second intervals as well): Sep 29 17:04:38 hostname avahi-daemon[506]: Registering new address record for IPV6-address on enp0s4.*. Sep 29 17:04:38 hostname avahi-daemon[506]: Withdrawing address record for IPV6-address on enp0s4. The IPV6 address in the Fedora logs is always the same. If I catch the dhcpv6.script running on the router using ps, the command it is running is always: {dhcpv6.script} /bin/sh /lib/netifd/dhcpv6.script eth0.2 ra-updated Is this normal/expected? If not, what data would I collect to determine if it's a configuration error, an ISP issue, or something that needs to be addressed in odhcp6c/netifd? Thanks -- Nathan ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel