[GE users] failed assumedly before job:can't write script file

Simon Gao gao at schrodinger.com
Fri Apr 21 18:40:22 BST 2006


    [ 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. ]

Reuti wrote:
> Hi,
>
> Am 20.04.2006 um 23:14 schrieb Simon Gao:
>
>> Hi,
>>
>> I have one user's job repeatedly failed on our SGE cluster with
>> following error:
>>
>> job 180271 caused action: Queue "all.q at node-0-11.local" set to ERROR
>>  User        = jim
>>  Queue       = all.q at node-0-11.local
>>  Host        = node-0-11.local
>>  Start Time  = <unknown>
>>  End Time    = <unknown>
>> failed assumedly before job:can't write script file
>> "job_scripts/180271" wrote only -1 of 3371433 bytes: Bad address
>>
>>
>> Also this user's jobs walked over each available node and caused ERROR
>> state and then moved on to next one. Other users don't any problem
>> running jobs from the same nodes.
>
> on the one hand it looks like an file permission problem, but usually
> the scripts are copied to the nodes with the settings of the admin
> user of SGE and are owned by him. Anyway we can have a look on it:
> which platform are you on, and has user jim the same settings/groups
> as the other users, for which it's working?
>
> -- Reuti
>
>
It turns out that the problem was caused by using qsub with a binary
directly. If wrap the binary in a script, then it worked without
problem. Now the question is can a user submit a binary command directly
using qsub, like "qsub <binary>"?

According to man page, it's ok to submit a binary using qsub. Is this a
bug with SGE 6.06u? or the man page is wrong?

Simon

=======================
SYNTAX
       qsub [ options ] [ command | -- [ command_args ]]

       qsh [ options ] [ -- xterm_args ]

       qlogin [ options ]

       qrsh [ options ] [ command [ command_args ]]

       qalter [ options ] wc_job_range_list [ -- [ command_args ]]

       qalter [ options ] -u user_list | -uall [ -- [ command_args ]]

       qresub [ options ] job_id_list

DESCRIPTION
       Qsub  submits  batch jobs to the Grid Engine queuing system. Grid
Engine sup-
       ports single- and multiple-node jobs. Command can be a path to a
binary or  a     <====
       script  (see -b below) which contains the commands to be run by
the job using

=======================




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