> This should hopefully help:
>
> https://wiki.italiangrid.it/twiki/bin/view/CREAM/SystemAdministratorGuideForEMI2#3_13_Sharing_of_the_CREAM_sandbo
This wiki only talks about sandbox staging in and out, but doesn't talk about where the job is actually run. If the area is shared between CE and WN then that doesn't mean I want 5000 jobs to all cruch i/o on that same folder over an NFS server or what not. What I want is the area for in/out to be shared, but the job execution should happen in a scratch area, not in /home/ that is let's say a NFS volume or glusterfs or what not that is not designed to handle such IOPS/BW as 5000+ jobs create.
So right now we have:
CREAM sandbox -> WN /home/user/xxxx -> job execution -> stageout
How I would like this to be:
CREAM sandbox -> WN /home/user/xx -> move stuff to /scratch/xxxx -> job execution -> stageout -> clean up /scratch
Mario Kadastik, PhD
Researcher
---
"Physics is like sex, sure it may have practical reasons, but that's not why we do it"
-- Richard P. Feynman
|