CAUTION: The timings are approximate and the order may change
Invitations to the Zoom meeting will be sent separately - but PLEASE register via this Indico web page to receive the details.
Notes by Dave K. (2 July 2020)
|Shawn Mc Kee|
Apologies: Andrea Sciaba, Kars Ohrenberg
1. DaveK welcomed all to the meeting - replacing the normal 2-day F2F meeting at CERN - cancelled because of COVID-19.
Agenda was agreed.
2. Tier-1/LHCOPN/LHCONE status - BrunoH/EdoardoM
Tier1 - nothing changed. No news from Russian T1; FNAL T1 - FTS server still IPv4-only.
Bruno reported that he had proposed IPv6-only LHCOPN to their recent meeting - willing to consider at an appropriate time (when IPv4 traffic is low).
Edoardo showed LHCOPN traffic stats. We looked at the top-10 talkers over IPv4. Agreed that trying to understand why these transfers happen over IPv4 would be useful. DuncanR says he will investigate.
3. IPv6 topics in IETF - TimC
An nice and extensive tour of many of the current issues in IPv6 inside IETF - see slides for details.
Lots of emails inside IETF about Segment Routing and packet steering wars!
4. IPv6 at Imperial - David Stockdale
See slides. Very interesting presentation about building an IPv6-only cluster and problems encountered.
Problems included: OpenVPN support for IPv6 on MACs; PBS batch problems; lost of booting problems; issues with Jumbo frames - see slides for more details.
--- coffee break ----
5. IPv6 and NET-WG subgroups - Shawn McKee
See slides. A very interesting review of the sub-groups being setup right now (people welcome to join), and the areas where use of IPv6 may be very useful.
We (the working group) clearly need to keep in touch with these activities.
6. Roundtable updates
DuncanR reported on current status of Kubernetes and IPv6/Dual-stack both in general and at CERN (see slides). Goal is to do some testing with CERN and ATLAS when possible.
RajaN reports that a problem of running LHCb work on IPv6-only CPU at Brunel had been failing because of problems with the Dirac portal. This had been investigated (with Raul at Brunel and others) and quickly solved/fixed.
QMUL now has dual-stack worker nodes on whole cluster.
7. AOB and Next meetings
Next one hour meeting agreed for Thursday 2nd July starting at 16:00 CEST.
Agreed that no F2F meeting at CERN in September. We will hold another virtual meeting (date to be set during the 2nd July meeting). Agreed that today has been very useful - 2 hours was too short - we should aim for 2 hours 30 minutes at the next virtual meeting.
Meeting closed - 25 minutes overrun - apologies!