IPv6 BGP segfault with bird 1.4.5
Hi, recently bird6 has been crashing quite often (several times a week) on one of our systems (I don't know why only one of two systems with very similar config is affected). Our config contains ~50 BGP peerings, the number of exchanged routes is about 100 in total. I was able to get a core of the crash, it can be found at http://home.universe-factory.net/neoraider/bird6.core I've put the corresponding binary (with debug information) at http://home.universe-factory.net/neoraider/bird6 bird was built on Arch Linux with GCC 4.9.2. Please let me know if you need any more information... Thanks in advance, Matthias
On Mon, Jan 12, 2015 at 11:02:10PM +0100, Matthias Schiffer wrote:
Hi, recently bird6 has been crashing quite often (several times a week) on one of our systems (I don't know why only one of two systems with very similar config is affected). Our config contains ~50 BGP peerings, the number of exchanged routes is about 100 in total.
Please let me know if you need any more information...
Hi, could you send me your config file and log fife? -- Elen sila lumenn' omentielvo Ondrej 'Santiago' Zajicek (email: santiago@crfreenet.org) OpenPGP encrypted e-mails preferred (KeyID 0x11DEADC3, wwwkeys.pgp.net) "To err is human -- to blame it on a computer is even more so."
On 01/13/2015 01:45 AM, Ondrej Zajicek wrote:
On Mon, Jan 12, 2015 at 11:02:10PM +0100, Matthias Schiffer wrote:
Hi, recently bird6 has been crashing quite often (several times a week) on one of our systems (I don't know why only one of two systems with very similar config is affected). Our config contains ~50 BGP peerings, the number of exchanged routes is about 100 in total.
Please let me know if you need any more information...
Hi, could you send me your config file and log fife?
I've attached the configs and what I could get of our logs from before the crash. Matthias
On Tue, Jan 13, 2015 at 02:02:31AM +0100, Matthias Schiffer wrote:
On 01/13/2015 01:45 AM, Ondrej Zajicek wrote:
On Mon, Jan 12, 2015 at 11:02:10PM +0100, Matthias Schiffer wrote:
Hi, recently bird6 has been crashing quite often (several times a week) on one of our systems (I don't know why only one of two systems with very similar config is affected). Our config contains ~50 BGP peerings, the number of exchanged routes is about 100 in total.
Please let me know if you need any more information...
Hi, could you send me your config file and log fife?
I've attached the configs and what I could get of our logs from before the crash.
I guess the problem manifests because of two protocols sharing IP address: protocol bgp ff_nrw3 from bgp_icvpn { neighbor fec0::a:cf:0:55 as 65530; } protocol bgp ff_westpfalz1 from bgp_icvpn { neighbor fec0::a:cf:0:55 as 65242; } You could try attached patch, it should fix the probable source of problems. -- Elen sila lumenn' omentielvo Ondrej 'Santiago' Zajicek (email: santiago@crfreenet.org) OpenPGP encrypted e-mails preferred (KeyID 0x11DEADC3, wwwkeys.pgp.net) "To err is human -- to blame it on a computer is even more so."
On 01/13/2015 02:47 AM, Ondrej Zajicek wrote:
On Tue, Jan 13, 2015 at 02:02:31AM +0100, Matthias Schiffer wrote:
On 01/13/2015 01:45 AM, Ondrej Zajicek wrote:
On Mon, Jan 12, 2015 at 11:02:10PM +0100, Matthias Schiffer wrote:
Hi, recently bird6 has been crashing quite often (several times a week) on one of our systems (I don't know why only one of two systems with very similar config is affected). Our config contains ~50 BGP peerings, the number of exchanged routes is about 100 in total.
Please let me know if you need any more information...
Hi, could you send me your config file and log fife?
I've attached the configs and what I could get of our logs from before the crash.
I guess the problem manifests because of two protocols sharing IP address:
protocol bgp ff_nrw3 from bgp_icvpn { neighbor fec0::a:cf:0:55 as 65530; }
protocol bgp ff_westpfalz1 from bgp_icvpn { neighbor fec0::a:cf:0:55 as 65242; }
You could try attached patch, it should fix the probable source of problems.
Thanks for the quick reply. I've applied the patch and will continue to run bird with the broken config for a while to see if the crash occurs again.
Hi, We've just encountered a very similar issue, in a DFZ setup. The router has a number of transit and peering peers. We lost the layer 2 connectivity for most of these peers, which resulted in bird6 crashing when it detected that the first peer was unreachable: Jan 20 04:53:07 estran bird6: lyonix_sfr: Error: Hold timer expired Jan 20 04:53:07 estran kernel: [208450.426908] bird6[5735]: segfault at 18 ip 00007ff06e5f6aa4 sp 00007fffd5dd0fe0 error 4 in bird6[7ff06e5eb000+74000] At the same time, bird correctly walked over all unreachable peers and didn't crash: Jan 20 04:53:14 estran bird: lyonix_sfr: Error: Hold timer expired Jan 20 04:53:25 estran bird: ipmax1: Error: Hold timer expired Jan 20 04:53:51 estran bird: lyonix_rs2: Error: Hold timer expired Jan 20 04:54:04 estran bird: ielo: Error: Hold timer expired This is on Debian wheezy, with the bird package from wheezy-backports (Version : 1.4.5-1~bpo70+1) BGP config is attached. There isn't anything in the logs, other than the two lines above. Any idea? We have three peerings in the same VLAN, but unlike Matthias, the neighbours addresses are different. On Mon, Jan 12, 2015 at 11:02:10PM +0100, Matthias Schiffer wrote:
Hi, recently bird6 has been crashing quite often (several times a week) on one of our systems (I don't know why only one of two systems with very similar config is affected). Our config contains ~50 BGP peerings, the number of exchanged routes is about 100 in total.
I was able to get a core of the crash, it can be found at
http://home.universe-factory.net/neoraider/bird6.core
I've put the corresponding binary (with debug information) at
http://home.universe-factory.net/neoraider/bird6
bird was built on Arch Linux with GCC 4.9.2.
Please let me know if you need any more information...
Thanks in advance, Matthias
participants (3)
-
Baptiste Jonglez -
Matthias Schiffer -
Ondrej Zajicek