GNOME Bugzilla – Bug 668286
IPv6 default route log errors
Last modified: 2012-03-22 14:24:25 UTC
Created attachment 205641 [details] head -n 1000 daemon.log.1 | grep -i networkmanager > daemon.grep.log When using a dual stack network, errors.log, daemons.log and messages.log files are flooded with errors. I attached errors.log from the Arch Linux bug tracker, where this bug is also filed. https://bugs.archlinux.org/task/27281?string=IPv6+default+route+log+errors&project=1&type%5B0%5D=&sev%5B0%5D=&pri%5B0%5D=&due%5B0%5D=&reported%5B0%5D=&cat%5B0%5D=&status%5B0%5D=open&percent%5B0%5D=&opened=&dev=&closed=&duedatefrom=&duedateto=&changedfrom=&changedto=&openedfrom=&openedto=&closedfrom=&closedto=
Debian also has this bug: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=649205 Log files with this bug are pretty annoying
Still alive in 0.9.3.997. Spam /var/log/errors.log For example : Mar 21 13:32:46 localhost dhcpcd[794]: eth0: sendmsg: Cannot assign requested address Mar 21 13:32:53 localhost NetworkManager[717]: <error> [1332333173.30466] [nm-system.c:1061] nm_system_replace_default_ip6_route(): (eth0): failed to set IPv6 default route: -1 Mar 21 13:33:09 localhost pulseaudio[1356]: [pulseaudio] pid.c: Daemon already running. Mar 21 13:55:28 localhost kernel: [ 4.830336] data transfer fail --- Mar 21 13:55:28 localhost /usr/sbin/crond[677]: (CRON) INFO (Syslog will be used instead of sendmail.): Aucun fichier ou dossier de ce type Mar 21 13:55:30 localhost dhcpcd[783]: eth0: sendmsg: Cannot assign requested address Mar 21 13:55:34 localhost NetworkManager[729]: <error> [1332334534.445357] [nm-system.c:1085] nm_system_replace_default_ip6_route(): (eth0): failed to set IPv6 default route: -14 Mar 21 13:55:36 localhost NetworkManager[729]: <error> [1332334536.414775] [nm-system.c:1085] nm_system_replace_default_ip6_route(): (eth0): failed to set IPv6 default route: -14 Mar 21 13:55:57 localhost pulseaudio[1331]: [pulseaudio] pid.c: Daemon already running. Mar 21 14:06:38 localhost kernel: [ 678.949931] INFO: rcu_preempt detected stalls on CPUs/tasks: {} (detected by 0, t=18002 jiffies) Mar 21 14:06:38 localhost kernel: [ 678.949932] INFO: Stall ended before state dump start Mar 21 14:09:38 localhost kernel: [ 859.056580] INFO: rcu_preempt detected stalls on CPUs/tasks: {} (detected by 0, t=72034 jiffies) And spamming continues every 3 minutes until I killed networkmanager :(
The fix for this is: diff --git a/src/nm-system.c b/src/nm-system.c index e15b85b..fad687f 100644 --- a/src/nm-system.c +++ b/src/nm-system.c @@ -1041,7 +1041,7 @@ replace_default_ip6_route (int ifindex, const struct in6_addr *gw) g_return_val_if_fail (route != NULL, -ENOMEM); /* Add the new default route */ - err = nm_netlink_route6_add (route, NULL, 0, gw, NLM_F_REPLACE); + err = nm_netlink_route6_add (route, &in6addr_any, 0, gw, NLM_F_REPLACE); if (err == -NLE_EXIST) { /* FIXME: even though we use NLM_F_REPLACE the kernel won't replace * the route if it's the same. Should try to remove it first, then but I believe this is a libnl difference between 1 and 3. Before I commit I need to check and see if this fix will work for libnl1 as well. Is everyone who's experiencing the error using libnl 3.x?
> but I believe this is a libnl difference between 1 and 3. Before I commit I > need to check and see if this fix will work for libnl1 as well. Is everyone > who's experiencing the error using libnl 3.x? I have this with Ubuntu 12.04, which uses libnl 3.2.3. Debian and Arch have libnl 3.2.7
Fix works on libnl1 as well. Pushed as 71ad568cfb5392d63e2899b5122829ba55fd5f80. We should no longer get log messages for error -14 (NLE_MISSING_ATTR) when attempting to set the IPv6 default route.