Skip to content

QOS - Quality of Service

Topola - qos

[username@hpc ~]$ sacctmgr show qos format=name,MaxWall,Priority,GrpTRES,MaxJobs%15,MaxSubmitJobsPerUser,MaxTRESPU%15,MaxTRESPA%25
      Name     MaxWall   Priority       GrpTRES         MaxJobs MaxSubmitPU       MaxTRESPU                 MaxTRESPA 
---------- ----------- ---------- ------------- --------------- ----------- --------------- ------------------------- 
    normal                      0                                                                                     
       hpc  7-00:00:00        750                            64         128                                           
    plgrid  7-00:00:00        750                            64         128                                           
       egi  7-00:00:00        500                                                                                     
     meteo                   1500   

Okeanos - qos

okeanos-login1 /home/username> sacctmgr show qos format=name,MaxWall,Priority,GrpTRES,MaxJobs%15,MaxSubmitJobsPerUser,MaxTRESPU%15,MaxTRESPA%25
      Name     MaxWall   Priority       GrpTRES         MaxJobs MaxSubmitPU       MaxTRESPU                 MaxTRESPA 
---------- ----------- ---------- ------------- --------------- ----------- --------------- ------------------------- 
    normal                      0                                                                                     
       hpc  2-00:00:00        750                                        32                                   node=90 
  internal  2-00:00:00       2000       node=64                                                                       
     meteo  2-00:00:00       1000      node=138   

RYSY - qos

username@rysy ~ $ sacctmgr show qos format=name%20,MaxWall,Priority,MaxTRESPU%60
                Name     MaxWall   Priority                                                    MaxTRESPU 
-------------------- ----------- ---------- ------------------------------------------------------------ 
              normal  2-00:00:00        100                                                       node=2 
               short    00:15:00      10000                                                              
                prio  2-00:00:00      10000                                                              
                long  7-00:00:00          1                    cpu=8,gres/gpu=1,gres/ve=1,mem=90G,node=1 
                none                      0                                                        cpu=0 
       internal_test  7-00:00:00          0                                             gres/ve=8,node=1 
             special  4-00:00:00          0                                                              
                 low                     10                                                              
   limit_interactive    12:00:00       1000     cpu=8,gres/cpuonly=8,gres/gpu=1,gres/ve=1,mem=90G,node=1 

Abbreviations:

  • TRES - Trackable RESources. A TRES is a resource that can be tracked for usage or used to enforce limits against

  • PU - Processing Unit.

Ta strona używa plików cookies.
Polityka Prywatności    AKCEPTUJĘ