Add option to delay propegating SIGINT to child process

[jhesketh]

Fixes #1497

Ctrl+C sends a SIGTERM to both tox and the running process, causing the
running process to receive the signal twice once tox passes it along.
This can often cancel or interfer with the process's cleanup.

Instead, add an option to allow a process to suicide before sending the
SIGTERM.

Co-Authored-By: Stefan H. Holek stefan@epy.co.at

Also includes a minor fix for correcting the location of the documentation for the existing timeout options: The interrupt_timeout and terminate_timeout options are set on the environment, not globally.

Contribution checklist:

• wrote descriptive pull request text
• added/updated test(s)
• updated/extended the documentation
• added relevant issue keyword in message body
• added news fragment in changelog folder
  • fragment name: <issue number>.<type>.rst for example (588.bugfix.rst)
  • <type> is must be one of bugfix, feature, deprecation,breaking, doc, misc
  • if PR has no issue: consider creating one first or change it to the PR number after creating the PR
  • "sign" fragment with "by :user:<your username>"
  • please use full sentences with correct case and punctuation, for example: "Fix issue with non-ascii contents in doctest text files - by :user:superuser."
  • also see examples
• added yourself to CONTRIBUTORS (preserving alphabetical order)

[gaborbernat]

Why move these?

[jhesketh]

This PR bundles up two commits. I can separate them out if needed, but it is explained in the earlier message:

Move timeout config options into correct section

The interrupt_timeout and terminate_timeout options are set on the
environment, not globally.

[jhesketh]

(I'll update the PR message at least)

[jhesketh]

So the failing test (test_parallel_interrupt) is marked as flaky, and will fail (even without my patch) if you put a sleep at the beginning of Action.handle_interrupt. That likely explains the flakiness, but I am unsure where in the test it is expecting things to be cleaned up so quickly (adding sleeps or checking the processes has not helped in my debugging, so I'm probably missing something).

[gaborbernat]

Sorry can't investigate this myself at the moment, and until the tests are passing we can't go ahead with this. Tets are passing on master.

[jheld]

Are the master tests consistently passing?

[gaborbernat]

Almost always.

[jheld]

I'm confused. The docs suggest that tox already supports something like this.

interrupt_timeout, terminate_timeout.

Is that not what's being asked for?

[jhesketh]

I'm confused. The docs suggest that tox already supports something like this.

interrupt_timeout, terminate_timeout.

Is that not what's being asked for?

The problem is that Ctrl+C will send a SIGINT to the foreground processes. This means that tox receives it along with pytest (or whatever it may be). Because tox then sends a SIGINT to the child processes it receives it twice. This goes unnoticed in most cases, but it often means that while pytest is cleaning up it receives another SIGINT and then aborts the cleanup.

This patch gives the running process a chance to exit before sending a SIGINT. If it hasn't exited, it'll receive a SIGINT from tox (which may be the first one it has received, or the second).

[jhesketh]

So I've set the default suicide timeout to 0.

This ensures that the behaviour has not changed from previously, and that programs get (at least one) SIGINT immediately. This also fixes the flaky test that appear to be sensitive to timing.