Support 'canceling' remote job status.
A status of canceling
, as is planned to be introduced, will indicate to Runner that the execution of the job should enter a canceling
state, rather than immediately abort everything.
What this effectively means is that a status of canceled
doesn't execute after_script
, but a status of canceling
does - however, we'll handle this logic in a separate MR (!2350 (closed))
This MR is concerned with the acceptance of the new status and how it propagates through the system.
To achieve this, this MR:
- We change
trace.Cancel
totrace.Abort
. This will be used for completely terminating a build. - We reintro
trace.Cancel
, but with a new meaning. This will be used to terminate the build stages, but allows graceful build stages to still execute (such asafter_script
- the only 'graceful' stage we plan on supporting in this first iteration).
Why was this MR needed?
What's the best way to test this MR?
What are the relevant issue numbers?
Edited by Steve Xuereb