HEPiX IPv6 working group meeting

Europe/Zurich
Vidyo

Vidyo

David Kelsey (Science and Technology Facilities Council STFC (GB))
Description

The timings are approximate!

HEPiX IPv6 WG call of the 11th of April 2019

Presents: Edoardo, Duncan, Bruno, Raja, Francesco, Andrea, Kashif, Martin
Excused: Dave

Updates

Francesco: the dcache patch was sent to KIT for testing, After testing Francesco should make a pull request that the dcache devels can accept or reject. Bruno will follow up.

Andrea: NtR

Bruno: ALICE is still in process of getting IPv6 ready. They have done all the systems except the relay host.

Duncan:
- Looking at IPv6 only WN in Brunel. some progress on xrootd monitoring
- Globus-connect is IPv4 only (kind of web FTS). It's used by some HPC sites in the US.

Edoardo:
LHCONE/OPN traffic seen at CERN has increased to 38% of the total https://twiki.cern.ch/twiki/bin/view/LHCOPN/LHCOPNEv4v6Traffic

Kashif: NtR

Martin: since end of January, people have complained about ipv6 packet loss from the Tier1 going out to JANET. The problem may be on the JANET router because the LHCOPN links to CERN are not affected. It's being investigated but not at top priority. The packet loss jumped up around the end of January. IPv4 is clean, so it can't be a physical problem.

Raja: LHCb moved from Castor to Echo so it's now dual-stack. In the next two weeks the other storage elements will be moved to dual-stack. https://ggus.eu/?mode=ticket_info&ticket_id=140456
Raja thanked the WG for the support.


Monitoring and Tier1/2 updates

Bruno: NtR

Andrea: FNAL is ready to switch on IPv6 on FTS

Andrea: slow progresses on IPv6 adoption at Tier2s. Closed only 2 tickets. 60.3% is done. Slight increase in the fraction of dual-stack storage for ALICE and ATLAS.
https://twiki.cern.ch/twiki/bin/view/LCG/WlcgIpv6


Suggestions for agenda items

Francesco: Francesco thinks that the document of F.Gont on IPv6 security (https://www.internetsociety.org/resources/deploy360/ipv6/security/ipv4-engineers) was disturbing. It's kind of assuming that people is not taking IPv6 seriously. For example RA guard was suggested by Cisco. The document says it can be bypassed with certain fragmented packets. In the past it would have been considered a bug, but now rfc 7113 says that RA guards can be evaded and doesn't suggest to fix it. Instead of fixing it, the RFC suggests to not use it. The whole document and IETF kind of assume that many things in IPv6 are broken and there is no urge to fix them. What do other people think about it?  It could be discussed at the meeting.

Bruno: abstract for CHEP19

 

 

 

There are minutes attached to this event. Show them.
    • 16:00 16:05
      Agenda review 5m
    • 16:05 16:20
      Updates - any urgent news? 15m

      Including sites and experiments

    • 16:20 16:45
      Monitoring updates, Tier 1 & Tier 2 status 25m
      Speakers: Dr Andrea Sciabà (CERN), Bruno Heinrich Hoeft (KIT - Karlsruhe Institute of Technology (DE))
    • 16:45 16:55
      Suggestions for agenda items - F2F meeting CERN 2/3 May 2019 10m
    • 16:55 17:00
      AOB and dates of next meetings 5m