We have seen it mostly with bwa jobs. Following is a query that I used to find 
out job/tools having output dataset size zero. I am not a database expert, so 
any suggestions for improving this query are welcome. :)

{{{
SELECT j.id, j.tool_id, j.tool_version, j.create_time, d.file_size, d.id 
FROM job as j, job_to_output_dataset, history_dataset_association, dataset as d
WHERE j.id=job_to_output_dataset.job_id AND 
job_to_output_dataset.dataset_id=history_dataset_association.id 
AND history_dataset_association.dataset_id=d.id AND d.file_size=0 
ORDER BY j.id desc;
}}}

I checked for corresponding filesystem (dataset_id.dat) size for each 
dataset.id and it seemed to be a problem with bwa_wrapper tool in most cases 
(only two exceptions - join1 and convert_characters1). However, it is not 
failing for all bwa_wrapper jobs. I have CCd here to Curtis who experienced 
this problem on our local install. He may have more details about particular 
job/tool involved in these cases. 

Thanks,
Shantanu. 


On Jul 12, 2011, at 9:33 AM, Kanwei Li wrote:

> Hi Shantanu,
> 
> Thanks for the report. Is there any pattern to the files being marked
> as empty? Certain filetype or histories?
> 
> -K
> 
> On Mon, Jul 11, 2011 at 3:07 PM, Shantanu Pavgi <pa...@uab.edu> wrote:
>> 
>> Hi,
>> 
>> In our local galaxy install (currently revision 50e249442c5a) we are seeing 
>> few output datasets as 'empty' in the galaxy web interface. The actual file 
>> size on the filesystem is ~ 4GB. These 'empty' files can be downloaded by 
>> clicking on download/view icons, however they can't be imported in other 
>> workflows. I looked into the database and figured out that file_size column 
>> value for corresponding dataset id is zero. I am guessing that's why it is 
>> being shown as empty in the web interface. Any suggestions on how to resolve 
>> this will be helpful.
>> 
>> --
>> Thanks,
>> Shantanu.
>> ___________________________________________________________
>> Please keep all replies on the list by using "reply all"
>> in your mail client.  To manage your subscriptions to this
>> and other Galaxy lists, please use the interface at:
>> 
>>  http://lists.bx.psu.edu/
>> 


___________________________________________________________
Please keep all replies on the list by using "reply all"
in your mail client.  To manage your subscriptions to this
and other Galaxy lists, please use the interface at:

  http://lists.bx.psu.edu/

Reply via email to