Project

General

Profile

18-July-2016

Attending: Bruno, Paola, Enrique, Vlad, Alex, Cristiana, Felipe, Dan, Paul

Computing issues

New limits on job lifetime is a pain. We were counting on a bug of the system to run, which is not available anymore. To bring back this system, we need to modify NGU. Alex won’t have much time so he asked for volunteers. Expected lifetime to zero, and internally check for time for each file. This is essentially what we had before. There’s an example in Ken’s talk in the FIFE workshop slides.

We need a place to track bugs and features of NGU. There will be a dCache downtime. It won’t let you move stuff between scratch and persistent and vice versa so we don’t get into trouble for doing this again. Keep ups stopped for a day due to dCahe issue.

Tags

Dan is having problems to work using Jenkins, so it’s not working so well for him. Paul comments that the browser ticket expires, so one needs to remake it every 7 days. This is needed to work on Jenkins, which might be the reason why Dan hasn’t been able to access Jenkins.

No rush in making the analysis tag, so it’s worth waiting for issues to be sorted.

Samples

ND RHC ideal

Finished, draining had 14 files, but definitions yield numbers that are inconsistent. There are a lot of error in FTS, around 400. But file size mismatches are not usually an issue. Probably there are either orphans or duplicates, so Vlad will check

Up-mu

Cristiana reconstructed most of the sample due to problems with metadata. She’ll make a draining dataset. Some fraction of the files get the detector type wrong (unknown), which has to be fixed manually. It’s a bit worrying but we still don’t know the reason. She’s almost done. Sample for calibration group (CRY with new fibre model) will be next.

ND top-up

Tapes should start asap.

System improvements

Job-specific metadata (Adam): 8 items that people want to have in the metadata. He’s getting started but needs some time from Alex to be sure of the details.

Keep-ups

Problems due to running out of time. Modified configuration, expected lifetime longer. Apparently they work but we need a tool to check on the output. We should try POMS to see whether monitoring. Bruno will follow instructions on monitoring, and ask Paola if he doesn’t get anything.