Add Job Types

When a new kind of job is required for backend processing, it has to be done properly in order to ensure it is used, set up, and processed by NCA.

• Make sure there aren’t already existing jobs that do what you want! There are a lot of jobs in NCA already, and some are meant to be very generic, such as JobTypeRenameDir.
  • Read and make sure you understand all structs in src/jobs that implement Process
• Create a new JobType in src/models/job.go.
  • Add the JobType to the const list
    • Make sure the string is 100% unique within that list!
  • Add the new JobType to the ValidJobTypes list
• Create a new struct that implements the Process method.
  • Use an existing Go file if it makes sense (e.g., another metadata or filesystem job) or create a new one in src/jobs/.
  • Make sure you document the type! What is its purpose?
  • Need an example? The metadata jobs are very simple and can be found in src/jobs/metadata_jobs.go.
• Wire up the JobType to the concrete Process implementor
  • This is done in src/jobs/jobs.go, in the DBJobToProcessor function
• Queue a job of the new type.
  • See src/jobs/queue.go
  • You might need to create a new arg value in src/jobs/queue.go, like JobArgSource, JobArgWorkflowStep, etc.
  • You will certainly need to create the job and push it into a queue. This happens in a Queue... function (e.g., QueueBatchForDeletion).
• Make something run jobs of the new type.
  • For almost any new job, you’ll just add the type to an existing runner function in src/cmd/run-jobs/main.go (runAllQueues). This ensures a simple job runner invocation (with the watchall argument) will run your new job type.