Choose timezone
Your profile timezone:
Tape Challenge '22
→
Europe/Zurich
, , ,
joint tape challenge
-
Attendees : Lisa Paspalaki, Artur Darovic Gottmann, Andreu Pacheco Pages, Christophe Haen, Edoardo Martelli, Enrico Fattibene, Eric Fede, Haykuhi, Mario Lassnig, Oliver Keeble, Preslav Konstantinov, Shigeki, Xin Zhao
-
ATLAS : quick updates
-
CMS plan :
-
Wrapped up DT and A-DT weeks. Some follow-up tests with some sites this week.
-
Full report at next week meeting.
-
-
LHCb plan :
-
DT test last week : reached targets with many sites. NCBJ not ready, to be tested at another time. CNAF throughput high but failure rate also high. RAL had good success rate, but transfer rate was low.
-
A-DT test this week. FTS staging limits were not tuned properly, many sites got flooded with too many staging requests.
-
srm+https protocol was used in this challenge for LHCb, including RAL Antares with the https protocol.
-
-
Sites feedbacks
-
CERN : everything is fine. Will provide CTA plots on VO traffic in later report.
-
FNAL : tuned # of parallel streams, slow traffic was observed as streams scaled up, further study needed. FTS throughput plot shows lower rate than the actually tape staging rate, because some files already exist at destination, try to repeat the staging tests with no disk replicas pre-existent.
-
KIT : not issue with tape challenge but in production, observation is disk replica of files are preferred over tape replicas. KIT aggregates files from the same dataset on tapes, so all files from a dataset will be staged out to tape disk buffer anyway. Can Rucio use the tape disk buffer copy instead ? Mario — can tune “distance” setting in Rucio for this to work.
-
-
Next meetings:
-
Next week tape challenge meeting : VO reports
-
Post-mortem from sites will follow in following meetings.
-
Next week DOMA general meeting : report on performance results
-
April GDB meeting : report on more comprehensive post-mortems
-
There are minutes attached to this event.
Show them.
