chore(container-runner): default stop grace to 10s

[abcxff]

No description provided.

[abcxff]

Stack for rivet-dev/actors

• #5556 fix(container-runner): read cgroup v1 memory usage under gvi …
• #5548 feat(container-runner): log version and git sha at actor sta …
• #5549 feat(container-runner): setup instance memory and cpu usage …
• #5546 fix(rivetkit): keep actor event loop alive through shutdown …
• #5539 feat(container-runner): report actors as crashed on unexpect …
• #5536 feat(container-runner): log unexpected platform SIGTERM as a …
• #5530 fix(rivetkit-core): cancel driver alarm before sqlite teardo …
• #5531 feat(container-runner): keep instance warm instead of self-e …
• #5533 chore(container-runner): default stop grace to 10s 👈
• #5528 fix(rivetkit): log and surface actor startup failures
• #5558 fix(sdks): drop rivet-util metrics dep from protocol crates …
• main

Get stack: forklift get 5533
Push local edits: forklift submit
Merge when ready: forklift merge 5533

_{change xnpvnuup}

[claude]

@-

[claude]

Review

Small, well-scoped change: lowers the default stop_grace_secs from 25s to 10s and generalizes doc comments away from Cloud-Run-specific wording to platform-agnostic language (container-runner/src/main.rs, container-runner/src/child.rs). No logic changes beyond the default value.

Verified consistency:

• effective_stop_grace() (main.rs:161) takes min(stop_grace, signal_child_stop_grace()) on signal-initiated shutdown, where signal_child_stop_grace() is 40% of the 10s SIGTERM_BUDGET default (4s). The signal-shutdown path is unaffected by this change since 4s was already the binding constraint and remains so at 4 < 10.
• The engine-initiated stop path (actor.rs:151, child.stop(cfg.stop_grace)) now waits up to 10s instead of 25s before SIGKILL, which is the actual behavioral change from this PR.
• sleep_grace_period (main.rs:314) is derived as stop_grace + 10s, so it drops from 35s to 20s as a side effect. Worth double-checking this stays comfortably above real-world shutdown time for the game-server processes this runner hosts, since core aborts on_destroy mid-stop if this deadline is too short (per the comment above it).
• No stale 25 references remain in container-runner/ source itself.

Still out of sync (unchanged since last review):

• docs/content/docs/container-runner.mdx:81 (Configuration table, --stop-grace-secs row) still documents the default as 25 (| \--stop-grace-secs` | `RIVET_STOP_GRACE_SECS` | `25` | ... |`). This is the public-facing docs page for this flag and will actively mislead users about the runner's out-of-the-box SIGTERM-to-SIGKILL grace.
• container-runner/examples/unity-demo/Dockerfile:32 still has a comment stating (default 25).

Both should be updated to 10 in this PR — see the docs-sync reference (.claude/reference/docs-sync.md) for the general rule on keeping this table current alongside code defaults.

No other bugs, security, or test coverage concerns. This is otherwise a config-default/comment change with no new logic paths, and no new tests are needed for a default-value tweak like this.