fix(sd_notify): send Stopping on shutdown and use timeout/2 for watchdog interval

[mohammadp1001]

Summary

Follow-up to PR #322's post-merge review, addressing items 1, 2, and 3 from #325 in
cda-extra's systemd_notify implementation:

• Send NotifyState::Stopping to systemd when the shutdown signal fires, so an
  intentional shutdown isn't mistaken for a watchdog failure.
• Replace the timeout - 5s (floored to 1s) watchdog notify interval with
  timeout / 2, matching systemd's documented convention
  (sd_watchdog_enabled(3)).
  The previous formula collapsed to "always notify every 1s" for any realistic
  watchdog timeout, since .min(1s) was acting as a cap rather than a floor.
• Renamed determine_interval to watchdog_notify_interval (item 2.1) to make
  its purpose clearer, and updated the doc comments accordingly.
• Set default-features = false on the workspace sd-notify dependency in
  the root Cargo.toml. No functional change: sd-notify enables no
  features by default today.

Checklist

• I have tested my changes locally
• I have added or updated documentation
• I have linked related issues or discussions
• I have added or updated tests

Related

Follow-up to #322

[github-actions]

📊 Coverage Report (unit + DoIP/CAN/mixed integration)

Total line coverage: 74.3%

Full build artifacts

[github-actions]

⚠️ 2 CRAP regression(s) detected

↑ 2 regressed · ★ 1 new · ↔ 4 moved · ↓ 0 improved · 1707 unchanged · — 1 removed
✓ = clean, ▲ = moderate, ✗ = crappy; Δ = change since baseline; CC = cyclomatic complexity; Cov % = line coverage percentage

	CRAP	Δ	CC	Cov %	Function	Location
✓	4.4	+0.4	4	70.8	UdsManager::start_variant_detection_for_ecus	cda-comm-uds/src/variant.rs:43
✓	6.4	+0.1	6	78.6	UdsManager::mark_group_unreachable	cda-comm-uds/src/variant.rs:200
✓	2.0	NEW	1	—	watchdog_notify_interval	cda-extra/src/systemd_notify.rs:90

↔ 4 moved

	CRAP	Δ	CC	Cov %	Function	Location
✗	42.0		6	—	trigger_watchdog	cda-extra/src/systemd_notify.rs:101 ← cda-extra/src/systemd_notify.rs
▲	30.0		5	—	fold_status	cda-extra/src/systemd_notify.rs:128 ← cda-extra/src/systemd_notify.rs
▲	12.0		3	—	create_sd_notify_task	cda-extra/src/systemd_notify.rs:40 ← cda-extra/src/systemd_notify.rs
✓	2.0		1	—	fold_health_state	cda-extra/src/systemd_notify.rs:143 ← cda-extra/src/systemd_notify.rs

🔥 Top hot spots above threshold

	CRAP	CC	Cov %	Function	Location
✗	600.0	24	0.0	EcuManager::map_table_struct_from_uds	cda-core/src/diag_kernel/payload_decode.rs:670
✗	306.0	17	0.0	dop_variant_to_schema	cda-core/src/diag_kernel/schema.rs:410
✗	272.0	16	0.0	MbedtlsError::name	comm-mbedtls/mbedtls-rs/src/error.rs:67
✗	240.0	15	0.0	TlsStream::poll_read	comm-mbedtls/mbedtls-rs/src/async_stream.rs:381
✗	210.0	14	0.0	DoipDiagGateway::send_functional	cda-comm-doip/src/lib.rs:648
✗	210.0	14	0.0	UdsManager::send_functional_group	cda-comm-uds/src/functional_group.rs:193
✗	156.0	12	0.0	update_mdd_uncompressed	cda-database/src/mdd_data.rs:364
✗	156.0	12	0.0	has_new_artifacts	cda-storage/src/recovery.rs:265
✗	156.0	12	0.0	diag_service::functional_data_request	cda-sovd/src/sovd/functions/functional_groups/data.rs:243
✗	110.0	10	0.0	DiagServiceError::from	cda-comm-doip/src/connections.rs:73
✗	110.0	10	0.0	comparams::executions::id::get	cda-sovd/src/sovd/components/ecu/operations.rs:273
✗	90.0	9	0.0	params_to_schema	cda-core/src/diag_kernel/schema.rs:188
✗	90.0	9	0.0	add_phys_constr_range	cda-core/src/diag_kernel/schema.rs:306
✗	90.0	9	0.0	EcuManager::map_table_key_from_uds	cda-core/src/diag_kernel/payload_decode.rs:622
✗	90.0	9	0.0	establish_tls_ecu_connection	cda-comm-doip/src/ecu_connection.rs:406
✗	90.0	9	0.0	create_tls_stream	cda-comm-doip/src/ecu_connection.rs:501
✗	90.0	9	0.0	remove_descriptions_recursive	cda-sovd/src/sovd.rs:923
✗	85.5	16	35.2	service::executions::ecu_reset_handler	cda-sovd/src/sovd/components/ecu/operations.rs:1346
✗	72.0	8	0.0	TlsStream::async_handshake	comm-mbedtls/mbedtls-rs/src/async_stream.rs:237
✗	72.0	8	0.0	UdsManager::send_functional_to_gateway	cda-comm-uds/src/functional_group.rs:39
✗	72.0	8	0.0	UdsManager::set_functional_state	cda-comm-uds/src/functional_group.rs:398
✗	72.0	8	0.0	comparams::executions::id::put	cda-sovd/src/sovd/components/ecu/operations.rs:401
✗	72.0	8	0.0	generate_config_cmd	cda-main/src/lib.rs:213
✗	70.6	10	15.4	DiagDataValue::try_into	cda-core/src/diag_kernel.rs:211
✗	60.3	11	25.9	load_config_with_storage_override	cda-main/src/config.rs:147

…and 38 more, see CI artifact for the full report.

— 1 removed

• determine_interval (was 2.0) — cda-extra/src/systemd_notify.rs

[alexmohr]

suggestion: We should keep some safe guard, so Duration != Duration::Zero because tokio will panic when using Duration::Zero as tick interval

[alexmohr]

@mohammadp1001 Please do not resolve my threads yourself. I will resolve them if I agree with the solution. It helps tracking what I commented and what I reviewed, instead of going through the closed threads.

[mohammadp1001]

Hi, Anything else missing?

[alexmohr]

1ms interval will lead to considerable system load. It's legal from the systemd configuration as it supports intervals starting a 1 micro second. But imho it's not a sane configuration for the CDA, we should log a warning once.

[theswiftfox]

@mohammadp1001 can you check the pipeline failure and fix the reported clippy lint?

[mohammadp1001]

Sure, I will check.