Skip to main content

You're viewing an archived page. It is no longer being updated.

RIPE 71

Open Source Working Group Minutes - RIPE 73
Thursday, 19 November 2015, 11:00-12:30
WG Chairs: Ondrej Filip and Martin Winter
Scribe: Michael Frearson

A. Administrative Matters (Working Group Chairs)

The slides are available at:
https://ripe71.ripe.net/presentations/136-RIPE71-Opensource-WG-Agenda.pdf

The Working Group Chairs, Ondrej Filip and Martin Winter, welcomed the participants. It was announced that the minutes from the previous RIPE Meeting had not yet been distributed, so discussion of these minutes was not possible.

B. GoBGP: Yet Another OSS BGPD?

Ishida Wataru, NTT Japan

The slides are available at:
https://ripe71.ripe.net/presentations/135-RIPE71_GoBGP.pdf

Thomas King, DECIX, asked for a second implementation that scales to the same levels as Bird. He thanked Ishida for the presentation and expressed his positive anticipation of future collaboration.

Martin Winter asked Ishida about multi-protocol support (multi-protocol on the one BGP session). Ishida confirmed that IPv4 and IPv6 capability exchanging on the same session is possible with GoBGP.

C. FreeRouter - Networking Swiss Army Knife

Csaba Mate, Developer of FreeRouter, Network Engineer @niif/as1955

The slides are available at:
https://ripe71.ripe.net/presentations/21-p-ripe.pdf

An audience member asked Csaba how he deals with a garbage collector in Java and how it affects performance in the router. Csaba replied that he uses G1 garbage collector and that it's very fast. Csaba then gave a demonstration of its speed. Csaba went on to explain that using GCC could be an alternative to dealing with a garbage collector, but he had found the performance to drop by up to 50%, plus GCC does not use the extensions of modern CPUs.

The audience member then asked whether Csaba misses timers while dealing with protocols. Csaba replied that timers aren't an issue and in BGP the open message carries the hold timer, and from this he can calculate the other timers. Csaba then explained his process using OSPF.

D. Towards Secure Input Handling in C/C++ Code

Jan Včelák, Knot DNS Team Leader, CZ.NIC

The slides are available at:
https://ripe71.ripe.net/presentations/164-ripe71-oss-vcelak-secure-input-02.pdf

Robert Kisteleki, RIPE NCC, suggested that Jan could have shown an example bug he found, to demonstrate the operation of the code. Jan responded that if one were to take a look at the American Fuzzy Lookup (AFL) webpage, there's a list of projects and bugs that were discovered by AFL, and examples of the bugs found can be looked up there.

Shane Kerr, BII, said he was looking at Jan's approach for finding bugs where he would set a flag in context and check it later. Shane said it seems dangerous because if you refactor your code, you might not do the checks at the right place. Shane asked if there were any problems in usage that way.

Jan responded that this was done only recently – about six months ago – and that his impression is that the code is much easier to read. There's a lot of resource record types you have to parse, and there's a lot of code. If you want to review the code for potential problems you easily get distracted by an error check every second line. For now, Jan said, this approach helped, but there may be some problems further down the line.

E. OpenSource Lightning Updates

E1. Open Source Toolset to Interact with RIPE Atlas

Robert Kisteleki, RIPE NCC

The slides are available at:
https://ripe71.ripe.net/presentations/105-atlas-cli-oswg.pdf

E2. jAtlasX - Access RIPE Atlas Through Java

Sascha Bleidner, DE-CIX

The slides are available at:
https://ripe71.ripe.net/presentations/89-jAtlasX_Access_RIPE_Atlas_through_Java.pdf

E3. Current Status of OpenBSD/OpenBGPd

Peter Hessler, OpenBSD

The slides are available at:
https://ripe71.ripe.net/presentations/165-openbsd-status.pdf

Kaveh Ranjbar, RIPE NCC, put a question to the room: the RIPE NCC Executive Board has been investigating supporting some Open Source projects that are good for the community. Kaveh then said he suggested to Peter to submit a proposal to the Executive Board with regards funding for this project. Kaveh asked the audience to demonstrate if they think RIPE NCC should support these toolsets. There was a show of hands in favour. This concluded the presentation sessions.

Martin Winter then encouraged participants to talk to people at RIPE 71 about various Open Source tools.

Martin announced that there had been no candidates for the Chair positions of the Open Source Working Group, and that Martin and Ondrej would therefore continue as Working Group Chairs for another term.

Martin reminded the group to vote in the Programme Committee (PC) election.

Finally, Martin announced that there were still slots for the lightning talk sessions on Friday, and that anyone interested should contact the PC for further details.