Job Scheduler Library
The library exposes methods to:
- create a new job scheduler (in two different ways)
- start a job
- stop a job by its ID
- get the output of a job
- get the status
- wait for all jobs completion (this would be used only in static apps - like the example main - not in the server that’s already waiting for some other events).
There are two example applications:
- a worker, that's an example of how the library could be used to spin up child processes;
- a main app, that's showing:
- concurrent example jobs;
- how to run the scheduler (so, again, it starts a job, prints the output, prints the status, waits for completion, in a similar way as a server app);
- how to run child jobs (in the same way as the worker does).
Requirements
To test and run the project you need to have an environment configured with cgroup v1. The included Dockerfile is built upon Ubuntu 20.04 that can simplify this on Mac machines running Docker Desktop.
To spin that up, run
make docker-build
make docker-run
and then check that you have the required cgroup version with
mount | grep cgroup
Another possible setup is to run Virtualbox with the same Ubuntu 20.04 running on it.
Building the worker
The project contains an example of an executable that can answer to "child" commands.
It can be built with:
go build -o bin/ ./cmd/worker
Note: Some tests assume that this application is contained the bin folder.
Testing
Run the tests with
go test ./...
And check for race conditions with
go test ./... -race
Execution
Run the project with
go run . run EXECUTABLE ARGS
For example
go run . run ./test.sh 5 1