These are chat archives for Exa-Networks/exabgp

22nd
May 2017
Brian Johnson
@brijohn
May 22 2017 13:08
@thomas-mangin The commit 9ee5fbf243 (tweak how we parse/handle local-as auto) seems incorrect, auto_discovery was implemented to allow local_address to be auto discovered during the bind process for outgoing connections. It shouldn't have any effect on the local_as value, but with this patch when I set local-address to auto my ASN becomes 0 even though local-as is properly set.
Thomas Mangin
@thomas-mangin
May 22 2017 13:45
Sorry - I was trying to look at ASN not Address … Mixed things up this week-end …
Thomas Mangin
@thomas-mangin
May 22 2017 13:50
@brijohn should now be back to working - sorry - oops, me bad !
There is still some work to be done on the changes on peer.py ongoing ..
Brian Johnson
@brijohn
May 22 2017 13:53
Alright thanks. could some of that peering work involve the reason that keep alive messages do not seem to be sent? When i removed the above problem patch on my test copy i was able to see peers exchange open messages but no keepalives were then sent leaving them stuck in the OpenConfirm state.
Thomas Mangin
@thomas-mangin
May 22 2017 14:07
Yes, sorry this work was not supposed to be commited upstream only on my tree but my finger done a typo
I think I can fix the reason of the issue with one comment, let me check ..
Try now
@brijohn sorry for the pain … I try to not break master .. but clearly yesterday, I failed
On the + side .. 4.0.0 was released so you can use it :-)
Brian Johnson
@brijohn
May 22 2017 15:11
Yeah, saw the release of 4.0.0 the other day. Thanks for the fixes on master I'll test and make sure they seem to be working. I was attempting to test the new version of master to see if issue 411 is indeed fixed, but hit these other issues before I could even test it.
Thomas Mangin
@thomas-mangin
May 22 2017 15:17
411 is unfortunately not done yet (not hard to fix tho).
Brian Johnson
@brijohn
May 22 2017 15:25
Your comment in 411 says otherwise since you imply it has been fixed in current master. Is this also part of the peer refactoring that is not yet complete?
Brian Johnson
@brijohn
May 22 2017 16:38
@thomas-mangin Oh oops i just realized i got the issue number i was referring to wrong. i mean issue 441, sorry for the confusion there.
Thomas Mangin
@thomas-mangin
May 22 2017 16:41
Thank you for checking 441
Brian Johnson
@brijohn
May 22 2017 16:41
Also i tested your fix that was preventing peering from coming up. It does now claim to sent a keepalive following the open messages, but the peer is sending me the following notification back 'peer reset, message [notification received (5,0)] error[State machine error / Unspecific]'
Thomas Mangin
@thomas-mangin
May 22 2017 16:42
I will check this tonight all being fine but if you can send me an exabgp -d .. I may be able to have an “ahah” moment and fix this quickly now
Brian Johnson
@brijohn
May 22 2017 17:06
@thomas-mangin Whats the best way to send you a copy of the -d output for exabgp?
Thomas Mangin
@thomas-mangin
May 22 2017 17:29
mail at first @ last . com
Thomas Mangin
@thomas-mangin
May 22 2017 20:23
@brijohn the reactor is working again :-)