The ManageIQ application can run as a collection of worker processes, without a centralized server. As a developer it may be easier to start the worker that you are currently working on, rather than starting the whole server and letting it launch the worker for you.
As of https://github.com/ManageIQ/manageiq/pull/15669 this will be possible much more easily than it was before.
run_single_worker.rb script allows a single worker to be started without needing a server to monitor it. You simply provide the worker class and optionally a queue name and port as environment variables and the worker will start.
There are additional options available which can be seen in the script’s usage documentation.
Foreman is a simple process management tool which will start up a group of processes at the same time.
It handles things like spinning up multiple instances of a partucular process and even will increment the built in PORT environment variable for each worker and worker type.
An example procfile can be found in the root of the ManageIQ repo. This file illustrates some common ways to start different types of workers, including web server workers, queue workers, and provider based workers.
A developer should create a
Procfile which suits their particular needs then run
bin/rake evm:foreman:start. This task will configure the local server to look started and then start the processes in the
These processes will run in the console in which the rake task was run until the user kills them with a Ctrl-C.
When we start using Foreman, a server process is not started. This has some implications for how the application will work. Most notably, settings will not be refreshed for workers and roles will not be able to be changed at runtime.
If you need a worker to pick up a changed setting or role, just kill the process in the terminal and start the workers back up.
Parity with containers
As we move towards running workers in containers, this method will be the most similar invocation to the start command in one of these worker containers.
The goal in the future is to replace forked workers with ones that run in containers, so having a similar way to run the workers in development will be important.