>>>>> In <87vco15vc6...@tsathoggua.smtp.nada.kth.se> >>>>> Andreas Davour <dav...@pdc.kth.se> wrote: > >>>>> In <4f107b9a.6080...@schedmd.com> > >>>>> Danny Auble <d...@schedmd.com> wrote: > > Andreas, have you read through this yet?
> > http://www.schedmd.com/slurmdocs/priority_multifactor.html > > Depending on the number of users you have in the account you created 2 > > might not be much of a difference than everyone else's 1 when > > normalizing happens. > Ok, now I have re-read that part and I think I get it. I have > PriorityWeightFairshare=100000 in my slurm.conf but I'm a bit unsure > about the part the db play in this. > Do I also have to set priorities and weights to accounts and users > using sacctmgr? I'm not sure the documentation is unclear or if I have > just not thought enough while reading it. > Just to make it clear. Do I need to add entried in the db for > everything I want to factor in into the fairshare? Or do the schema > contain everything you can collect accounting data on? I'm not that > familiar with mysql so I'm not sure how to check. > If I have understood correctly, I not only have to set the > PriorityWeightFaishare directive in slurm.conf, but also add weight to > accounts, clusters and fairshare elements in the database, right? I realized that some examples might be a good idea, illustrating my problems. I think I have enetered stuff in the db using sacctmgr but it still looks wonky. This is how the weights are configured (why is PRIORITY empty?) b01c21n04# sprio -w JOBID PRIORITY AGE FAIRSHARE Weights 1000 100000 This is what I have entered using sacctmgr b01c21n04# sacctmgr list assoc tree format=cluster,account,user,fairshare | grep atls-021 ruth atlas atls-021 1000 b01c21n04# sacctmgr list assoc tree format=cluster,account,user,fairshare | grep atlp-000 ruth atlas atlp-000 1 b01c21n04# sacctmgr list assoc tree format=cluster,account,user,fairshare | grep atlm-000 ruth atlas atlm-000 20000 b01c21n04# sacctmgr list assoc tree format=cluster,account,user,fairshare | grep swegrid ruth klasm swegrid 1 Now when I look at the jobs in the queue, taking the newest. This is how it looks like. Am I misunderstanding something? This looks reversed to me. b01c21n04# sprio -ln -j 54368 JOBID USER PRIORITY AGE FAIRSHARE JOBSIZE PARTITION QOS 54368 atlp-000 0.00000000 0.0049876 0.0000000 0.0000000 0.0000000 0.0000000 b01c21n04# sprio -ln -j 54253 JOBID USER PRIORITY AGE FAIRSHARE JOBSIZE PARTITION QOS 54253 atls-021 0.00000055 0.0238128 0.0236545 0.0000000 0.0000000 0.0000000 b01c21n04# sprio -ln -j 53973 JOBID USER PRIORITY AGE FAIRSHARE JOBSIZE PARTITION QOS 53973 swegrid 0.00000343 0.0536847 0.1470953 0.0000000 0.0000000 0.0000000 b01c21n04# sprio -ln -j 52440 JOBID USER PRIORITY AGE FAIRSHARE JOBSIZE PARTITION QOS 52440 atlm-000 0.00000249 0.0551389 0.1064769 0.0000000 0.0000000 0.0000000 Or am I just using too small numbers? I expected atlm-000, job 52440, to have much higher FAIRSHARE number than job 53973?? /andreas -- Systems Engineer PDC Center for High Performance Computing CSC School of Computer Science and Communication KTH Royal Institute of Technology SE-100 44 Stockholm, Sweden Phone: 087906658 "A satellite, an earring, and a dust bunny are what made America great!"
