The modules provided by this package provide the glue between the RECAST Control Center and the analysis plugins.
In recastbackend/backendtasks.py a Celery shared_task provides a wrapper function around a user-specified python function entry point. This wrapper will execute setup and teardown actions before executing the entry point.
Currently a catalogue of existing plugins is hardcoded into recastbackend/catalogue.py, but it is foreseen to have this configurable.
Any analysis plugin VM must have this package insstalled in addition to the plugin code. Once the environment is prepared for analysis, one can start a Celery worker process via
celery worker -A recastbackend.fromenvapp:app -l info -Q <queue name>
to start accepting RECAST analysis jobs.
when connected to a docker daemon via DOCKER_HOST (like on a Mac with boot2docker)
docker volume create --name workdirsdata
export RECAST_IN_DOCKER_WORKDIRS_VOL=$(docker volume inspect workdirsdata|grep Mountpoint|awk '{print $NF}'|sed 's|\"||g')
docker-compose -f cap-compose.yml up -d && docker attach misc_headnode_1
In the head node you can the issue
recast-directsub cap http://physics.nyu.edu/~lh1132/dummycomplex.zip complex_analysis/fullworkflow.yml /recastdata/outhere --track
this will have the results appear in /recastdata/outhere
:
ls -lrt /recastdata/outhere