Hello,

Go home today to find that Boinc had crashed because the 16GB drive dedicated to it went 100% full (12K free) despite a 'leave at least' preference setting of 0.5GB. I extended the logical volume to 20GB and restarted the client, by time I could open the manager and check disk usage Rosetta had risen to 16.75GB usage. I forced a project update to clear the 'computational error' and 'ready to report' tasks which free about 50MB of the original 76 tasks I now have 1 'aborted by Project' and 60 'suspended by user' tasks. Of the 60 tasks 31 have recorded elapsed time. repeating the update cleared the 'aborted by project' entry but had no effect on the disk space used. Aborting the unstarted tasks took the usage to 16.62GB. So currently the average disk per work-unit is around 550MB.

So my question is: as a ballpark figure how much disk space should I expect a work-unit to consume?

If this is your system with 40 CPUs... I'd say 1GB per CPU would be a starting point.

---

Rosetta Moderator: Mod.Sense

Thanks,

Yes it is the 40 CPU system, 2 x Xeon E5-2630 v4 , 10 core a piece + hyperthreading; gets me by as a desktop :-). O.K. really it my virtualization host, BOINC keeps the basement warm when i'm not messing with DevOps, MongoDB or playing CIV.

Guess I've a weekend of LVM migrations and PV resizing ahead of me. Not Run BOINC on a dedicated volume before and clearly got the sizing wrong.