Configure Timeout per Task

[zirkelc]

Goals

1. No fails on CI because of hanging tasks
2. No CI timeouts after 30 minutes
3. Easier troubleshooting

Non-goals

No response

Background

Some tasks like serverless deploy do not handle errors correctly and instead of terminating the process, they report the error to the console and stop responding. These tasks must be terminated by the user with Ctrl+C.

I run multiple build tasks in CI sequentially and if a task hangs, it stops Turbo until the entire workflow is terminated by the GitHub Actions timeout.

While I agree that the root cause of the problem should be fixed, the hanging task, this is not possible in all cases so it would be useful to have a timeout for each task.

In my case this was related to this issue #8281 because I thought Turbo was hanging on CI, but I found out that it was hanging because a task was hanging and it was possible to see this on CI. Only when I ran the CI pipeline locally did I see that there was an error on a task.

Proposal

I suggest to add a timeout option to the config (globally or per task) to set the maximum time a task can use before it is terminated by Turbo.
A more sophisticated option would to reset the timeout after each output to stdout, so long running tasks won't get killed because they report their progress to the console.