Hi Susan, I would love to see the result of the most popular open source implementation testing with IXIAs, they are lovely boxes, just a bit too expensive for most of us.
Many of the home brewed implementations ExaBGP replaced had never seen one. If ExaBGP was tested against one, I was never told neither. But there is indeed hope as the few open source implementations I checked did the right thing when it came to check the OPEN size. Thomas Sent from my iPad > On 10 Mar 2017, at 21:59, Susan Hares <[email protected]> wrote: > > Thomas: > > Real world testing is really, really important. > > If Quagga, Bird, or GoBGP have been run against test devices (IXIA, etc), the > testers used to test against these cases. Most of these took their initial > code from open-source GateD which had this feature. The open-source GateD > was distributed into many different commercial code bases – there may be some > hope. > > Cheerily, Sue Hares > > From: Thomas Mangin [mailto:[email protected]] > Sent: Thursday, March 9, 2017 9:27 AM > To: [email protected] > Cc: Susan Hares; 'Thomas Mangin'; [email protected] > Subject: Re: [GROW] Do you want BGP to extend the message size for all BGP > messages or just UPDATES. > > Hi Nick, > > Looking at Quagga, Bird, GoBGP (I took a few minutes to read their code and I > am not overly familiar with any of them), they all seems to perform the 4096 > bytes check and send NOTIFICATION, so I would only expect badly home brewed > solution to suffer from the issue (not to say we should not care). > > That said, I would also assume the code path for this scenario has never been > run in production (and perhaps ever) in the life of any BGP implementation so > while it may look good, it may not do what is expected. I believe some "real" > world testing is in order .... > > Thomas > On 2017-03-09 12:20, Nick Hilliard wrote: > > Sue: I'd be cautious with your approach. First, it's not guaranteed > that some badly coded bgp stack wouldn't crash with a 4097 byte OPEN > message, and secondly, you're not guaranteed that just because the stack > supports 4097 bytes on open due to e.g. unintentional coding reasons, > that it actually supports 4097 bytes by design and that it actually > works properly.
_______________________________________________ GROW mailing list [email protected] https://www.ietf.org/mailman/listinfo/grow
