Choose timezone
Your profile timezone:
HEPiX IPv6 working group meeting
→
Europe/Zurich
Vidyo
Vidyo
HEPiX IPv6 working group meeting - Vidyo - Thursday 11 June 2015
Present: Alastair Dewhurst, Thomas Finnern, Kashif Hafeez, Thomas Hartmann, Bruno Hoeft, Tiju Idiculla, Dave Kelsey, Fernando Lopez Munoz, Ewan MacMahon, Edoardo Martelli, Raja Nandakumar, Kars Ohrenberg, Francesco Prelz, Ulf Tigerstedt, Ramiro Voicu.
Apologies: Jiri Chudoba, Tony Wildish.
Actions and ongoing issues
- DaveK. Ask for news from ALICE/CostinG. Ongoing
- AndreaS. Dual-stack SAM3 monitoring. Ongoing. AlastairD reports that in ATLAS something is dual-stack on SAM. AndreaS will follow up.
- DaveK. Security Best Practices document. Ongoing. To be done before the GridKa school in September.
- FrancescoP. Setup for XRootD testing. Has started by recompiling XRootD version 4.2.1. Will now test with ATLAS/FAX.
Roundtable updates
a) INFN. FrancescoP reports that he was at CNAF a few days ago to assist in setting up perfSONAR. BUT routing issues still not solved. He has started work on XRootD.
b) ATLAS. AlastairD reports that Oxford and Glasgow IPv6 test SpaceTokens for ATLAS have started receiving transfers. He asks about our policy for dual-stack storage services as would be good to test in production. Answer: fully agree! But needs to be carefully controlled by the WG so that experiments are fully aware of any changes.
Conclusion: the group mail list should be used to inform of any new production dual-stack services so all are informed.
It was also agreed that the IPv6-only SE at Oxford should be removed from ATLAS functional tests.
Alastair also notes that there is very little time to worry about IPv6 now that data taking has started (true for all experiments!).
c) KIT. BrunoH. Bruno reports some progress with LHCOPN in the USA (FNAL).
d) CERN. EdoardoM. No news.
e) PIC. FernandoLM reports that he is now in contact with the Enstore developers working on an IPv6 version but there is no estimated date yet. They have finished changes with their new firewall. The offices network has been made dual-stack. During the next two weeks they plan to enable ATLAS dCache pools in IPv6.
f) DESY. KarsO reports that DFN has now enabled IPv6 on LHCONE for both DESY and KIT. DESY has also enabled eduROAM on IPv6 with a few hundred users per day.
g) RAL. TijuI reports that he is working on a Ceph/S3 storage gateway setup.
h) LHCb. RajaN reports he has revisited the current list of products on the group IPv6 application database and notes that LFC is no longer used. Dirac v6r13 has been formally released and is now being tested. The CERN fix to the broken (for IPv6) library has not yet been done.
i) DESY. ThomasF has nothing new to report.
j) KIT. ThomasH reports that FTS is still not working properly. UlfT had helped with the debug when it was concluded that gsiftp was falling back to PASV. It is not yet certain what is causing this but is correlated in time with an update to gfal2 but there are also ongoing issues with dCache. Thomas will downgrade to the version which worked in March and check what breaks the transfers.
k) Nordic. UlfT reports that they still have one tape system to fix and then they will be ready to move to dual-stack. There was a national IPv6 day earlier this week when all mobile networks enabled IPv6 (and has been left on). They went from 10k to 4M devices overnight! Next steps will be ADSL and Cable. He is experiencing dCache headnode reboot problems where randomly IPv6 does not get enabled. Perhaps some timing issue with Ubuntu?
l) email before the meeting from JiriC at Prague. The addition of the IPv6 VO to the production SE golias100.farm.particle.cz is not yet finished.
m) email before the meeting from QMUL (Dan Traynor). There was an issue with the Atlas SAM test at QMUL when we setup our one of our cream CE as dual stack IPV6 node. Note that it was passing the standard nagios ops tests at the time. Marian Babik at CERN has constrained the production SAM ATLAS production system to now be IPv4 only. We have also setup our storm SEs as dual stack IPV6. The ipv6.hepix.org VO is available on se02.esc.qmul.ac.uk and se01.esc.qmul.ac.uk test SEs.
Testing plans
First step is to fix FTS on the testbed. AlastairD asks again as to why we need the testbed and why don't we move straight to production testing. DaveK says he would at least like to understand the FTS failure reasons and get things working there before moving to production. We also don't want to break services during data taking.
Should we test FTS somewhere else? We could ask Imperial? (note added after the meeting - eventually the FTS problems at KIT were traced to an upgrade in the Globus libraries - this has been reported).
We should discuss plans for testing on production at the next meeting.
Plans for another IPv6 workshop event
There seems to be no suitable pre-GDB slot during rest of 2015. We could do something at HEPiX/BNL in October but necessary experts may not be able to attend. It does not seem appropriate for something at the January/Feb WLCG Workshop. DaveK will discuss with Michel Jouvin (chair of GDB).
Dates of future meetings
Vidyo conference Thursday 9th July 2015 at 16:00-17:00 CEST
F2F meeting at CERN Tuesday/Wednesday 15/16 September 2015 at CERN (lunchtime to lunchtime).
Notes by Dave Kelsey
9 July 2015
There are minutes attached to this event.
Show them.