In past did a little and seem to remember they would run like 6-12 hours and upload, but now had 2 on 2 octos, who would not move after 100 hours Elapsed, about 10 seconds CPU time or an NCI usage of about 0.001%. The reason of returning was to build some WCG-MCM statistics for a new CPU.

Activated the trickle_debug, if maybe intermediate reports were going up, but none. There was though every so many hours a connect to WUPRop which then said requesting 0.00 seconds work and then being backed off for a bit.

When reading other threads about missing reported hours, went to look in my account, unawares this was being logged and was surprised only DSFL showed for 992 hours. I'd expected there to be more sciences / hours, not so though.

Anyway, the "I wonder" button was activated. It's very low CPU usage, the information is useful, but this way not much of anything in contribution.

P.S. 16 cores, 100 hours would have been 1600 hours of stats.

---

WUProp work units should last for exactly three hours before finishing and uploading, automatically downloading a new one. If yours are taking several hundred hours there is clearly an issue. Aborting a work unit might allow it to upload without losing any stats, but don't take my word on that.

Rob, your systems are hidden; we can't see what app versions were being used, the runtime...
Did you delete the WU's, try to reset the project,
or detach, delete project folders and then re-attach?

3h runtime is specified by the WU. Lots of people are contributing to the stats, but there have been problems with some app versions.

http://wuprop.boinc-af.org/results/projet.py?projet=World+Community+Grid&application=Mapping+Cancer+Markers