This mapping is not automatic - you need to write a small Python method to take these parameters specified by the user and map them to your cluster parameters. These methods are called dynamic job destinations and described on the wiki at: https://wiki.galaxyproject.org/Admin/Config/Jobs#Dynamic_Destination_Mapping If your method takes in a function keyword argument called "resource_params", Galaxy will build a dictionary from the user supplied parameters and send them to your function. So in your case {"memory1": 300} or something like that - and the method should build a destination with a native specification that uses this information. Hope this helps. -John On Fri, Mar 18, 2016 at 5:45 PM, Stephan Oepen <oe@ifi.uio.no> wrote:
hallo again, fellow galaxy users and developers,
as an extension to my original query, i am now wondering how the parameters in ‘job_resource_params_conf.xml’ map onto SLURM options? for example, i assume <param ... name="processors" ...> maps onto something like ‘--ntasks’ (or maybe ‘--ntasks-per-node’). are the ‘name’ values in the definition of job resource parameters standard keys defined for DRMAA, and drmaa-python know how to map these into SLURM parameters? or is there an explicit specification of that mapping somewhere?
we have succeeded in establishing per-tool defaults by putting these into the ‘nativeSpecifiation’ of multiple variants of the DRMAA destination. but now we would also like to customize the valid range and initial value that is presented to users when they decide to use the ‘custom’ job resource form in the tool configuration dialogue. in other words, we would like to do something like the following in ‘job_resource_params_conf.xml’
<param label="Memory" name="memory1" type="integer" size="2" min="1" max="16" value="1" ... /> <param label="Memory" name="memory4" type="integer" size="2" min="4" max="24" value="4" ... /> <param label="Memory" name="memory6" type="integer" size="2" min="6" max="24" value="6" ... />
and then associate a specific memory parameter with individual tools in ‘job_conf.xml’. but for that to work, i would have to understand the mapping to SLURM options and make it so that ‘memory1’ to ‘memory6’ all map to ‘--mem’ (or maybe ‘--mme-per-cpu’).
once i understand things better, i would of course be happy to contribute a summary for the galaxy wiki. for all i can see, current documentation does not cover job configuration and job resources in full detail.
with thanks in advance, oe
On Sun, Mar 13, 2016 at 2:31 PM, Stephan Oepen <oe@ifi.uio.no> wrote:
many thanks for taking the time to answer my query, gildas!
In your job_conf.xml, you can set per tool a destination.
i had realized that much (sending some of our tools to SLURM, running others on the local node), but i had failed to realize that one can of course have /multiple/ SLURM destinations, which all send to the same cluster but differ in their default resource parameters.
thanks again, oe
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: https://lists.galaxyproject.org/
To search Galaxy mailing lists use the unified search at: http://galaxyproject.org/search/mailinglists/