Last modified 2 years ago Last modified on 06/04/19 15:00:00

dCache Tier I meeting June 4, 2019

[part of a series of meetings]

Present, Paul), IN2P3(), Sara(), Triumf(), BNL(), NDGF(), PIC(), KIT(Xavier), Fermi(), CERN()


(see box on the other side)

Site reports


Things are going smoothly -- not a whole lot of change compared to last week.

Test transfers failing

CMS test transfers, ran at KIT with xrootd are randomly failing.

KIT opened a support ticket with SLAC xrootd (9648) complained about random failures.

With some investigation, they discovered that these failures were due to the 20 second timeout.

The test file is only 240 MiB, so 20 seconds should be easily satisfied; however, they have discovered that the transfer can take up to 8 minutes to complete.

For now, they have taken out the timeout from the test, allowing it to proceed.

The cause of these very slow transfer rates is currently not known, although Xavier believes that dCache is not the cause of the problem.

All the test transfers are all using IPv6. KIT currently has no by-pass for the firewall, so all IPv6 transfers are going through the firewall, which may be overwhelmed if there are lots of transfers concurrently.

It could also be a problem with the underlying block device.

Xavier is investigating.

Support tickets for discussion

[Items are added here automagically]


Same time, next week.