moq-boy: exit non-zero on reconnect give-up instead of hanging

[kixelated]

Problem

moq-boy gives up reconnecting to the relay after the mandatory 5m backoff timeout (added in #1443), but the process doesn't exit — it hangs as a zombie. This defeats systemd's Restart=always, so a relay outage (or any future deploy/relay rebuild) longer than the reconnect window silently kills the games and they never come back.

Root cause

The emulator runs on a spawn_blocking thread that loops forever (rs/moq-boy/src/main.rs:274). When the reconnect loop gives up, run() returns Err, main returns it, and the tokio runtime is dropped. Dropping the runtime blocks indefinitely joining that never-ending blocking thread, so the process hangs instead of exiting. The ctrl_c branch already side-stepped this with an explicit std::process::exit(0); the give-up path did not.

Fix

Capture the select! result and exit explicitly via std::process::exit (0 on clean shutdown, 1 on error) instead of returning from main. process::exit terminates immediately without dropping the runtime, so it can't hang on the blocking thread, and a real non-zero exit code reaches systemd. The give-up error is logged on the way out.

The 5m timeout is kept as-is (a permanent relay outage still surfaces as a non-zero exit rather than silently spinning forever), so Restart=always now correctly restarts the service.

moq-cli

No change needed. moq-cli has no spawn_blocking task, so returning Err from main already exits non-zero cleanly. It shares the same 5m give-up shape but never happened to hit the window.

Test plan

• cargo check -p moq-boy
• cargo clippy -p moq-boy (via nix toolchain)

(Written by Claude)

[coderabbitai]

Caution

Review failed

The pull request is closed.

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 2437c185-014d-42b8-99a4-5288c00a1a1c

📥 Commits

Reviewing files that changed from the base of the PR and between f5a4ffe and b58e454.

📒 Files selected for processing (1)

• rs/moq-boy/src/main.rs

---

Walkthrough

The PR changes rs/moq-boy/src/main.rs so the top-level tokio::select! result is captured into a variable. The Ctrl-C branch now returns Ok(()) instead of calling std::process::exit directly. After the select!, main matches the result: it logs and exits with code 1 on error, or exits with code 0 on success, preventing implicit runtime drop while background tasks run.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately summarizes the main change: fixing moq-boy to exit with a non-zero code when reconnect attempts are exhausted, rather than hanging.
Description check	✅ Passed	The description is comprehensive and directly related to the changeset, clearly explaining the problem, root cause, fix, and testing approach.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

✨ Simplify code

• Create PR with simplified code
• Commit simplified code in branch claude/hopeful-tharp-6af1b9

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@rs/moq-boy/src/main.rs`:
- Around line 449-453: The select arm handling Ctrl-C currently swallows any
io::Error by mapping tokio::signal::ctrl_c() to Ok(()); update the
tokio::select! arms (the block that awaits run(&config), jemalloc, and
tokio::signal::ctrl_c()) to propagate the Err from tokio::signal::ctrl_c()
instead of treating it as a clean shutdown — replace the `_ =
tokio::signal::ctrl_c() => Ok(())` arm with one that returns the Result from
tokio::signal::ctrl_c() (and add context as needed) so failures registering the
ctrl-c handler bubble up alongside errors from run and jemalloc.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 72411194-69f1-4686-a06f-a2f658fe9e3d

📥 Commits

Reviewing files that changed from the base of the PR and between b7a08ff and f5a4ffe.

📒 Files selected for processing (1)

• rs/moq-boy/src/main.rs