[GE users] All queues dropped because of overload or full

Reuti reuti at staff.uni-marburg.de
Thu Dec 13 21:03:35 GMT 2007


Am 13.12.2007 um 17:47 schrieb Alexandre Racine:

> Thanks for the Hyperthreading hint. I think I'll remove it in the  
> near future since most of processes do use 100% cpu. I'll test with  
> something heavy and compare.
>
>
> Getting back to the "All queues ..." message.
>
> I can see the statistician currently having only a few jobs now,  
> and still having the message. Infos below. Could the message still  
> be there since the load is higher then 1.75, witch is the default  
> in the queue configuration "load_thresholds       np_load_avg=1.75"?

no, no_load_avg is per core, hence with 8 cores you can put an  
absolute load of 14 on it. If you define slots=cores you can even  
remove it completely (i.e. set it to "none").

> I hope I am not asking too much questions :)
>
>
> $ qstat -j 186
> [...]
> scheduling info:            queue instance  
> "all.q at wasabi01.statgen.local" dropped because it is full

Seems okay: used slots=0, defined slots=0.

-- Reuti

>
>
> $ qstat -f
> queuename                      qtype used/tot. load_avg  
> arch          states
> ---------------------------------------------------------------------- 
> ------
> all.q at PAPRIKA                  BIP   3/14      4.84     lx24-amd64
>     167 0.55500 All_RLS_Me asseling     r     12/12/2007  
> 15:40:07     1
>     180 0.55500 pprd-sw_sn asseling     r     12/12/2007  
> 16:45:40     1 3
>     186 0.55500 rls-pbat35 asseling     r     12/13/2007  
> 11:35:48     1 19
> ---------------------------------------------------------------------- 
> ------
> all.q at oregano.statgen.local    BIP   1/8       2.32     lx24-amd64
>     166 0.55500 SIME_RLS_M asseling     r     12/12/2007  
> 15:40:07     1
> ---------------------------------------------------------------------- 
> ------
> all.q at wasabi01.statgen.local   BIP   0/0       0.67     lx24-amd64
>
>
> $ qhost
> HOSTNAME                ARCH         NCPU  LOAD  MEMTOT  MEMUSE   
> SWAPTO  SWAPUS
> ---------------------------------------------------------------------- 
> ---------
> global                  -               -     -       -        
> -       -       -
> PAPRIKA                 lx24-amd64     16  4.84   30.4G    2.7G     
> 1.9G     0.0
> oregano                 lx24-amd64      8  2.32   15.7G    4.3G     
> 1.9G     0.0
> wasabi01                lx24-amd64      8  0.67   14.6G  485.5M     
> 2.0G     0.0
>
>
>
>
>
>
>
>
>
> -----Original Message-----
> From: Reuti [mailto:reuti at staff.uni-marburg.de]
> Sent: Thu 2007-12-13 10:57
> To: users at gridengine.sunsource.net
> Subject: Re: [GE users] All queues dropped because of overload or full
>
> Am 13.12.2007 um 15:53 schrieb Alexandre Racine:
>
>> Oops, sorry I did confuse -conf and -msconf. So in -msconf it is
>> like this :
>>    schedd_job_info                   true
>>
>> So it is already to true.
>>
>> Currently my statisticians are doing some jobs and I have again the
>> same message.
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
> For additional commands, e-mail: users-help at gridengine.sunsource.net

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net




More information about the gridengine-users mailing list