I'm trying to set up a cloud based tool execution function where a job is started on a local galaxy server, but it is executed on a pulsar server that is hosted on a cloud server. So far I am able to start a job if the cloud server is running, but the aim is to be able to start the pulsar server when the job is started, and then shut it down once the job is finished. The pulsar server is hosted on an openstack cloud server, so my idea is to inject the openstack start command before the job is executed on the pulsar server. But before we move any further, perhaps this is already supported by some feature that has been implemented already? If not, then here is an outline of the idea I'm thinking of:
Start job on galaxy -> start openstack server with pulsar running on it -> execute the job -> shut down the openstack server when the job is finished -> done.
I got the idea of putting the openstack command in the tool xml file, but that would try to execute it on the remote server that I'm trying to start. And since the job_conf.xml file has the tool configured to be executed on the pulsar server, I need to somehow sidestep this configuration and execute the openstack locally but the job command remotely.
Is this possible in a clean way?