Hi,

Am 04.02.2014 um 17:32 schrieb Mark Dixon:

> I think this is a bug, but would like to hear other opinions :)
> 
> Over the years, we've run various versions of SGE and SoGE with a share tree 
> policy. I think it's always been the case that jobs in an error state still 
> attract tickets from the share tree policy - despite the fact that the job 
> isn't eligible to run.

This is for sure a matter of discussion. One could also argue, that because of 
the E state the admin should take proper action and remove the condition which 
caused the error. After this, the job should get the priority it deserves to 
instantly. If it is a bigger problem, the admin could put the job on hold. And 
this puzzles me more: although the job is put on hold, it still keeps the 
priority it once got.

For other waiting jobs it's working: attaching and removing the hold will 
recalculate the priority every time.

-- Reuti


> e.g.
> 
> $ qstat -ext -s p | head -5
> job-ID  prior   ntckts  name       user         project          department 
> state cpu        mem     io      tckts ovrts otckt ftckt stckt share queue    
>                       slots ja-task-ID 
> ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
>      1 0.50050 0.50000 job        bob          DEF              defaultdep 
> Eqw                              137170     0     0     0 137170 0.59         
>                            16
>      2 0.39808 0.39758 job        sue          DEF              defaultdep qw 
>                               46411     0     0     0 46411 0.06              
>                       50
>      3 0.34129 0.34079 job        sue          DEF              defaultdep qw 
>                               39781     0     0     0 39781 0.05              
>                       50
> 
> This means there are fewer tickets available for the scheduler to distribute 
> to other waiting jobs. In the extreme case, it could mean other pending jobs 
> have the same priority when they shouldn't.
> 
> Given that ticket allocations are calculated afresh every scheduling 
> interval, I don't think there's any point in errored jobs attracting tickets 
> like this.
> 
> Is that right?
> 
> Mark
> -- 
> -----------------------------------------------------------------
> Mark Dixon                       Email    : [email protected]
> HPC/Grid Systems Support         Tel (int): 35429
> Information Systems Services     Tel (ext): +44(0)113 343 5429
> University of Leeds, LS2 9JT, UK
> -----------------------------------------------------------------
> _______________________________________________
> users mailing list
> [email protected]
> https://gridengine.org/mailman/listinfo/users


_______________________________________________
users mailing list
[email protected]
https://gridengine.org/mailman/listinfo/users

Reply via email to