These are chat archives for Exa-Networks/exabgp

27th
May 2015
sanjmonkey
@sanjmonkey
May 27 2015 13:26
Hi TM! Thank you for picking up #277. I spent some time last night trying to see why its not caught in reserved
Thomas Mangin
@thomas-mangin
May 27 2015 13:59
I am looking into it now - should have a fix for you soon
Thomas Mangin
@thomas-mangin
May 27 2015 14:05
@sanjmonkey Hi Sandy, I have a patch for you on master, but I am not sure it is the reason for the session failing. You are trying to use Multi-Session it may be why.
See comment on Github
the patch is adding a “return self” so it does not print None - I could not get the establishment to fail like you have.
sanjmonkey
@sanjmonkey
May 27 2015 14:07
@thomas-mangin seen the change, cloning now
sanjmonkey
@sanjmonkey
May 27 2015 14:12
working!
Thomas Mangin
@thomas-mangin
May 27 2015 14:17
weird - you must have pulled a version of master when I was adding the draft-hostname support and was buggy !
sanjmonkey
@sanjmonkey
May 27 2015 14:17
thanks @thomas-mangin, it looks good… i saw you adding the draft-hostname and did try from your personal branch too
where i think the issue still exists
interesting choice of test peer BTW :smile:
Thomas Mangin
@thomas-mangin
May 27 2015 14:19
I left the code in but the capability is now removed
For the support - I just wrote it as people on the IETF list were starting to upset me with really buggy argument.
sanjmonkey
@sanjmonkey
May 27 2015 14:22
yes i saw! it proves its not a hard thing to implement if you did it in 1hr anyway!
Thomas Mangin
@thomas-mangin
May 27 2015 14:24
I had to add 40 minutes to fix left and right issues and I done it in a rush .. but yes, it was really easy
I wanted to be able to send the patch on the list within the hour ..
Now we will see how the draft evolve ( or not )
sanjmonkey
@sanjmonkey
May 27 2015 14:26
lol. we will measure you on features/hour.
Thomas Mangin
@thomas-mangin
May 27 2015 14:27
I do agree that there is issue with the length of the packet ( the capability max size is 255 but the max cap size could have been 1+255+1+255) but as no one spotted it I passed
Having only the hostname with a size of 255 would pass
if you add 513 bytes to a maximum of 4096 ( minus BGP headers ) it is indeed a big overhead
Thomas Mangin
@thomas-mangin
May 27 2015 14:42
You know that I am working on the CLI and a way to offload the BGP message parsing to another process/core. The core was modified for it and the offloading works ( but I would not call it production ready )
sanjmonkey
@sanjmonkey
May 27 2015 14:47
yes i did see you made lots of progress on CLI. when do you think you will be releasing 3.5.?
Thomas Mangin
@thomas-mangin
May 27 2015 15:07
I am at least a month away
Thomas Mangin
@thomas-mangin
May 27 2015 16:06
I like to mess with people head : thomas-mangin/exabgp@37fc126
Thomas Mangin
@thomas-mangin
May 27 2015 18:01
@sanjmonkey just to make sure .. are you running master in prod .. as it is currently going through large changes and code refactoring . 3.4.11 is the stable code ATM.
or 3.4 stable with today’s fix
sanjmonkey
@sanjmonkey
May 27 2015 18:49
Prod is a bit of a strong word! :) I have safety mechanisms in place with to prevent anything bad happening, but appreciate the advice
Thomas Mangin
@thomas-mangin
May 27 2015 19:26
I prefer to let you know that I am likely to break think in the next few weeks in advance than after the fact :wink2:
sanjmonkey
@sanjmonkey
May 27 2015 19:34
I know your plans ;) Defensively, I autovivify your json and search for keys I think should be there :P