On Tue, 13 Sep 2016, Martin Winter wrote:
Paul,
I’ve started it, but already found some BGP issues…
So somehow my CI testbed had a bug and when the basic setup of the BGP IPv6
tests failed (and caused all tests to skip),
then this was ignored (skipped instead of failed tests). Sorry :-(
This broke with the CI Pull Request #38 (and was fine up to and including
#37)
https://github.com/opensourcerouting/quagga/pull/38
Should have been flagged as failed and not as good…
The issue (on IPv6 BGP):
“No supported combination found in the received BGP4 Open
Message”
This works for plain IPv4 BGP, but fails for IPv6 sessions.
Weird. I don't see what in the 2 leak fixes of #38 could cause a
problem. I doubt your test even hits the paths either one change (you
don't go in and delete the peer in bgpd via the CLI interface do you?
and you're not testing AFI_ETHER either and anyway that path only is hit
when shutting down bgpd)?
Can you get the tcpdump?
Additionally, for AS4 BGP sessions (IPv4 or IPv6), there is a collision issue
which then closes the session:
(I thought you fixed this - did the fix only apply to 2-byte AS?)
No, generally. Collision window should be lower, and a possible bug in
handling was fixed.
However, that doesn't mean a collision is impossible - it's just
dependent on timing and can happen every now and then. The test needs to
be prepared for a retry. Either the test itself should run again, or the
wrapper around the test should try again if the fail was due to
connection collision. ??
From the bgpd log for the collision:
2016/09/13 18:41:04 BGP: [Event] Make dummy peer structure until read Open
packet
2016/09/13 18:41:04 BGP: 192.168.1.1 [FSM] TCP_connection_open
(Active->OpenSent)
2016/09/13 18:41:04 BGP: 192.168.1.1 passive open
2016/09/13 18:41:04 BGP: 192.168.1.1 sending OPEN, version 4, my as 500,
holdtime 180, id 192.168.1.101
2016/09/13 18:41:04 BGP: 192.168.1.1 send message type 1, length (incl.
header) 51
2016/09/13 18:41:04 BGP: 192.168.1.1 went from Active to OpenSent
2016/09/13 18:41:04 BGP: 192.168.1.1 rcv message type 1, length (excl.
header) 18
2016/09/13 18:41:04 BGP: 192.168.1.1 rcv OPEN, version 4, remote-as (in open)
23456, holdtime 90, id 192.168.1.1, inbound connection
2016/09/13 18:41:04 BGP: 192.168.1.1 [AS4] rcv OPEN w/ OPTION parameter len:
8, peeking for as4
2016/09/13 18:41:04 BGP: [AS4] found AS4 capability, about to parse
2016/09/13 18:41:04 BGP: 192.168.1.1 [AS4] about to set cap PEER_CAP_AS4_RCV,
got as4 65538
2016/09/13 18:41:04 BGP: %NOTIFICATION: sent to neighbor 192.168.1.1 6/7
(Cease/Connection collision resolution) 0 bytes
2016/09/13 18:41:04 BGP: 192.168.1.1 send message type 3, length (incl.
header) 21
2016/09/13 18:41:04 BGP: Notification sent to neighbor 192.168.1.1: type 6/7
Let me know if you need help to reproduce the issues
Scratching my head on the first one..
regards,
--
Paul Jakma | p...@jakma.org | @pjakma | Key ID: 0xD86BF79464A2FF6A
Fortune:
Humor in the Court:
Q: (Showing man picture.) That's you?
A: Yes, sir.
Q: And you were present when the picture was taken, right?___
Quagga-dev mailing list
Quagga-dev@lists.quagga.net
https://lists.quagga.net/mailman/listinfo/quagga-dev
