Async gitlab jobs, do we have to hack support for them ourselves?
Kick off a 1 hour CLI command that is just going to wait for 1 hour. Assume that after the 1 hour wait, something else needs to happen.
My ideas so far
- job create a pipeline schedule to poll for completion, then do the call back, call back then deletes the schedule when done.
- OR 24 hour heat beat to check for unresolved async jobs, when resolved run the callback.
Some day, yaml is going to have an `async` keyword.