HEPiX - IPv6 Meeting Notes (03 Sept. 20)

Chaired by Bruno Hoeft as DaveK not able to attend. Notes by Bruno too.

Present: Dimitrios Christidis, Edoardo Martelli, Francesco Prelz, Raja Nandakumar, Martin Bly

Apologized: Kars Ohrenberg, Tim Chown, Andrea Sciaba, Dave Kelsey

Roundtable:

Dimitrios Christidis:

• Tier-2 Desy-Zeuthen now IPv6 dual-stack ready

Edoardo Martelli:

• Proposal handed in to deploy the small data center of 8000 virtual images/nodes with public IPv6 addresses only

 Francesco Prelz:

• NTR

Raja Nandakumar:

• LHCb smoothly running
• At CERN VO-boxes with IPv4 only addresses

 Martin Bly:

• 100G link deployment to CERN are soon implemented
• Odd problem IPv4 transfers starting moderate but slowing down to a very low rate
  • Since traffic is running trough the firewall --> One suspicion is that the Firewall is the reason for the slowing down the transfer speed
  • IPv6 transfers are much faster (--> this is may be the carrot we where looking for)
  • Investigation is still ongoing

 Bruno:

• NTR

Tier1/Tier2 LHCOPN/LHCONE status:

Tier-1 : situation unchanged:

• FNAL : beginning of last month FTS traffic was increasing, but dropped down to more or less 0, e-mailed Phil DeMar, no response up to now
• RRC-KI-T1 - Kurchatov Institute in Moscow still IPv4 only

Tier-2 : Andrea updated his statistics (plots/tables)

https://twiki.cern.ch/twiki/bin/view/LCG/WlcgIpv6

• From June Desy-Zeuthen (DESY-ZN) is now dual-stack ready
• Vienne might have also done it, but I'm waiting for the confirmation of the experiments supported by it.

Tracking at DE-KIT of IPv4 top talkers:

Bruno showed plot (attached to agenda item).

During 4 hours inbound:

               Src IP Addr                                                                                  Bytes(%)

BNL:       192.12.15.221                                                                  1.1 T(39.5)
CNAF :   131.154.x.x (various IPv4 addresses)     221.6 G( 4.3)  
FNAL:    131.225.x.x (various IPv4 addresses)        90.9 G( 1.4)

 During 2 hours outbound:

CERN:   128.142.x.x  (various IPv4 addresses)        1.2 T( 32.5)

Can we do any IPv6-only testing now?

• Since CORONA many sites are more or less freezing all developing activities and only running services as it is there
• No activities are reported

September 2-day F2F (virtual) meeting:

• Take the IPv6-only testing on the agenda
  two test bed ideas:
  • Martin : set up a IPv6 only testbed at RAL (virtual infrastructure)
  • I will ask at DE-KIT for a view IPv6 only Workernodes (WN)

Do we continue to use Vidyo?:

• Was not explicitly raised at the meeting,
• But since CERN is moving away from Vidyo, i would suggest to follow CERN to ZOOM