Running jobs
Dexen clients
After creating a job definition and starting dexen, you can now starting running a job on Dexen. For this you will need a Dexen client. The client is the front end of Dexen that provides some way for users to interact with the server and nodes.
A variety of different clients could be developed. Currently, there is only one client, callled the CLI, or the Command Line Interface. This is a non-graphical client where you can type in types in commands. This client is designed for running multi-criteria optimisation problems. The client includes commands for generating snapshot graphs and pareto graphs.
A client with a Graphical User Interface (GUI) is also planned.
Workflow using the CLI
The key steps for running job on Dexen using the CLI are as follows:
- You submit the job to the server using the Dexen CLI. When submitting the job, you specify the folder on your local computer that contains the job definition. This folder will be zipped up and sent to the server.
- The server then runs your job.
- The server designates a node instance on which to run a master.
- The server sends the zipped package of the job definition to the master.
- The server assigns free slaves to the master periodically.
- The master then decides which task to schedule on the assigned slave.
- You can use your web browser to browse the job archive (described in more detail below).
- You can use the CLI to generate various files, including log files and pareto files. This will allow you to keep track of how the job is progressing.
Note that the CLI does not play any role in running the job. Once the job has been started, the CLI can be closed. Later on, a new CLI can be started (on the same computer, or a different computer) and can connect the the server running the job.
For any job, the CLI commands that produce the files need to have access to the data related to the job. The data can be downloaded from the server by issuing a download command. The data will be saved on the CLI's local hard disk. Closing the CLI will erase this data.
Job archive
For each job, the server creates a website that contains all the archived data for that job. Any type of data can be archived for each individual in the population. The job definition defines what type of data is archived.