Allow configuring the daemon inactivity timeout

[benaliabderrahmane]

Fixes #1239.

The daemon's inactivity timeout was hard-coded to 2 hours. We run robots for days with a large number of nodes, and once the daemon has timed out an occasional ros2 CLI call respawns a node and triggers DDS discovery, which shows up as a CPU spike. This makes the timeout configurable so a single long-lived daemon can be kept around instead.

• ros2 daemon start --timeout N shuts the daemon down after N seconds of inactivity. The default is still 2 hours, so behavior is unchanged unless you ask for it.
• A negative value disables the timeout, so the daemon stays up until it is explicitly stopped (same convention as helpers.wait_for).
• spawn_daemon() gains an inactivity_timeout argument; the auto-spawn path keeps the 2 hour default.
• Also fixes _ros2_daemon's existing --timeout: a negative value used to make the daemon exit immediately, now it disables the timeout.

This intentionally only affects daemons started explicitly; auto-spawned daemons keep the 2 hour default.

Added test/test_daemon_timeout.py covering the new argument and the short-timeout / never-timeout behavior.

[fujitatomoya]

just a minor comment. i think this optional argument is nice to have, i also think that use case is well explained.

[Copilot]

Pull request overview

This PR makes the ROS 2 CLI daemon’s inactivity shutdown behavior configurable, allowing long-lived daemons (including “never time out”) to avoid periodic respawns and DDS rediscovery spikes.

Changes:

• Exposes an inactivity --timeout option on ros2 daemon start (default remains 2 hours; negative disables the timeout).
• Plumbs the configured inactivity timeout through spawn_daemon() into the daemon server loop.
• Adds tests validating the CLI argument exposure and the timeout/never-timeout behaviors.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 3 comments.

File	Description
ros2cli/test/test_daemon_timeout.py	Adds new tests covering explicit daemon start inactivity timeout behavior.
ros2cli/ros2cli/verb/daemon/start.py	Adds ros2 daemon start --timeout and forwards it to spawn_daemon().
ros2cli/ros2cli/node/daemon.py	Extends spawn_daemon() with inactivity_timeout and passes it into the daemon server.
ros2cli/ros2cli/daemon/init.py	Implements “negative timeout disables inactivity timeout” behavior and updates _ros2_daemon --timeout help text.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[benaliabderrahmane]

Thanks for the review!

Pushed an update addressing everything:

• Moved the tests into test_ros2cli_daemon.py and dropped the separate file along with the duplicated fixtures, reusing the existing rmw isolation setup.
• Fixed the "the the" typo in the docstring (and the same one a few lines up in shutdown_daemon).
• Changed the timeout annotation on serve() / serve_and_close() to float, since a negative value gets turned into float('inf').
• Bumped the inactivity timeout in the shutdown test from 2s to 5s so it isn't racy on slower CI.

Kept it as a single commit since it's one logical change, happy to split it if you'd rather.

[fujitatomoya]

lgtm with green CI.

[fujitatomoya]

Pulls: #1240
Gist: https://gist.githubusercontent.com/fujitatomoya/be9302cd2ce538a423c908bc67759de2/raw/b4a0545a7d59f5dfa0906726ffebdac457064d5f/ros2.repos
BUILD args: --packages-above-and-dependencies ros2cli
TEST args: --packages-above ros2cli
ROS Distro: rolling
Job: ci_launcher
ci_launcher ran: https://ci.ros2.org/job/ci_launcher/19480

• Linux
• Linux-aarch64
• Linux-rhel
• Windows