[GE users] odd problem in new installation

Reuti reuti at staff.uni-marburg.de
Fri Dec 2 19:30:06 GMT 2005


Hi,

Am 02.12.2005 um 16:57 schrieb Josh Brandt:

>
> I had to reinstall gridengine (6.0u6) from source the other day,  
> and now I
> find that it's not working quite right.
>
> When jobs are submitted, they appear to run as they should, but  
> when they're
> done (and I've been testing this with a really simple "sleep 20"  
> script),
> they generate an error and throw the queue into error state,  
> stopping queue
> processing. The seemingly-finished jobs also don't then leave the  
> queue, but
> go back into pending jobs.
>
> qstat -j [jobnumber] has this error at the bottom:
>
> error reason    1:          12/02/2005 10:52:52 [0:410]: can't get
> configuration value for "enable_addgrp_kill"

which platform are you on? The ENABLE_ADDGRP_KILL seems to be a (for  
now) undocumented feature of the SGE configuration for the execd. You  
could try with:

qconf -mconf

to edit the SGE configuration and put there:

execd_params                 ENABLE_ADDGRP_KILL

and set it to none after you saved it afterwards again.

You are using the maintrunk for compilation and use in the whole  
cluster the same version? Your installation of SGE is shared to all  
nodes?

Cheers - Reuti


> I can't find anything about "enable_addgrp_kill" anywhere...
>
> Can anyone give me any idea of where this is going wrong?
>
> Josh
>
>
> -- 
> jbrandt at wpi.edu
> Senior Unix Systems Administrator
> Worcester Polytechnic Institute
>
> ---------------------------------------------------------------------
> 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