[GE users] consumables(licenses) and paralell environment not working...

Lönroth Erik erik.lonroth at scania.com
Wed May 16 17:00:55 BST 2007


    [ The following text is in the "iso-8859-1" character set. ]
    [ Your display is set for the "ISO-8859-10" character set.  ]
    [ Some special characters may be displayed incorrectly. ]

Just to clarify:

I need 1 licence (fluent_all) for the job fluent.
I also nedd 1 licence for each calculation process. So, if I run a job requesting 8 CPU:s, the total amount of licenses will be: 1 (fluent_all)+ 8 (fluent_par)

How is this best done?

/Erik

-------------



>>You are aware, that:

>>>> #$ -l fluent_all=1
>>>> #$ -l fluent_par=8

>>will be multiplied by 8, as you requested 8 slots?

>>-- Reuti

Oh! I didnt know that... How should I specify that I need:

1 fluent_all
8 fluent_par

In a correct way? Have I made a misstake in specifying 4 slots per que and "UNDEFINED" #processors in my que definition?

... Am I making any sense to you in my error?

Thanx for your input Reuti anyway, priceless!

/Erik

-----Original Message-----
From: Reuti [mailto:reuti at staff.uni-marburg.de] 
Sent: den 16 maj 2007 13:51
To: users at gridengine.sunsource.net
Subject: Re: [GE users] consumables(licenses) and paralell environment not working...


Am 16.05.2007 um 13:03 schrieb Lönroth Erik:

> Yes, they are.
>
> I have added the pe:s to the queues.
>
> I have 2 queues:
>
> short.ts102.q (all hosts from subcluster "ts102")
>    fluent_ts102_pe
>
> short.ts103.q (all hosts from subcluster "ts103")
>    fluent_ts103_pe
>
> This might be some other foobar thing I have done, maybve you can help 
> me point out a good error search path? Logfiles to look in etc.

You are aware, that:

>> #$ -l fluent_all=1
>> #$ -l fluent_par=8

will be multiplied by 8, as you requested 8 slots?

-- Reuti


>
> /Erik
>
> -----Original Message-----
> From: Reuti [mailto:reuti at staff.uni-marburg.de]
> Sent: den 16 maj 2007 12:34
> To: users at gridengine.sunsource.net
> Subject: Re: [GE users] consumables(licenses) and paralell environment 
> not working...
>
>
> Am 16.05.2007 um 09:05 schrieb Lönroth Erik:
>
>> My pe looks like this:
>>
>> # Version: 6.0u8
>> #
>> # DO NOT MODIFY THIS FILE MANUALLY!
>> #
>> pe_name           fluent_ts103_pe
>> slots             9999
>> user_lists        NONE
>> xuser_lists       NONE
>> start_proc_args   /opt/gridengine/mpi/myrinet/startmpi.sh -
>> catch_rsh $pe_hostfile /opt/mpich/myrinet_mx2g/gnu/bin/mpirun.ch_mx
>> stop_proc_args    /opt/gridengine/mpi/myrinet/stopmpi.sh
>> allocation_rule   $fill_up
>> control_slaves    FALSE
>> job_is_first_task FALSE
>> urgency_slots     min
>>
>> As you see I'm using myrinet_mx and apart from that I think this 
>> should work... My queues are setup with 4 slots and "UNDEFINED" 
>> processors.
>
> The PEs are also attached to the queues?
>
> -- Reuti
>
>
>> /Erik
>>
>>
>> -----Original Message-----
>> From: Dan.Templeton at Sun.COM [mailto:Dan.Templeton at Sun.COM]
>> Sent: den 15 maj 2007 16:51
>> To: users at gridengine.sunsource.net
>> Subject: Re: [GE users] consumables(licenses) and paralell 
>> environment not working...
>>
>>
>> Erik,
>>
>> The naive assumption would be that your PE has the slots attribute 
>> set to 0.
>>
>> Daniel
>>
>> Lönroth Erik wrote:
>>>
>>> I have problems getting jobs through in my SGE 6u8 environment.
>>>
>>> I have setup a license load_sensor as decribed by 
>>> http://bioteam.net/dag/sge-flexlm-integration/
>>> but when I submitt jobs and specify:
>>>
>>> #$ -l fluent_all=1
>>> #$ -l fluent_par=8
>>>
>>> I get from the scheduler:
>>>
>>> qstat -j 113
>>> ...
>>> ...
>>> ...
>>> script_file:                fluent.job
>>> parallel environment:  fluent_*_pe range: 8
>>> scheduling info:            cannot run in PE "fluent_ts102_pe"
>>> because
>>> it only offers 0 slots
>>>                             cannot run in PE "fluent_ts103_pe" 
>>> because it only offers 0 slots
>>>
>>> If I dont include the "-l fluent_xxx" resources, it works as normal.
>>>
>>> What am I doing wrong? I recall this has worked before, but now it 
>>> seems not to.
>>>
>>> /Erik
>>>
>>
>> ---------------------------------------------------------------------
>> 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
>>
>
> ---------------------------------------------------------------------
> 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
>

---------------------------------------------------------------------
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

---------------------------------------------------------------------
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