fix(tool): surface MessagePool service task errors in api_cmd test ctx

[0xDevNinja]

Summary of changes

Changes introduced in this pull request:

• Own the JoinSet in test_snapshot::ctx and generate_test_snapshot::ctx instead of passing &mut JoinSet::new() into MessagePool::new. The inline temporary was dropped on return, aborting every spawned mpool service task and silently swallowing any errors they emitted pre-cancellation.
• Move the owned JoinSet into a detached tokio::spawn(drain_mpool_services(...)) that drains it with the canonical while let Some(result) = services.join_next().await loop from daemon/mod.rs::propagate_error, logging errors and non-cancel panics via tracing::warn!.
• Add a shared drain_mpool_services helper in test_snapshot.rs (visibility pub(super)) and call it from both ctx() sites.

Detaching (vs. awaiting inline) is deliberate: the mpool services are long-lived, so an inline while let Some(...) would never return — that is precisely what caused codecov CI to time out and led to the earlier attempt being reverted from #6493. A detached drainer surfaces errors without blocking the caller.

Reference issue to close (if applicable)

Closes #6505

Other information and links

• Canonical drain pattern: src/daemon/mod.rs:842-851 (propagate_error).
• MessagePool::new signature confirmed at src/message_pool/msgpool/msg_pool.rs:~721 — takes services: &mut JoinSet<anyhow::Result<()>>, which matches the generic used here.
• Prior related PR (different file, same pattern): test: unit test for start_rpc #6493.
• No CHANGELOG.md entry — this is an internal test utility change and is not user-facing per the changelog guide at the top of that file.
• No new test: the fix is observability-only (previously-silent warnings now surface via tracing); asserting on log output would be flaky. Existing rpc_regression_tests_print_uncovered passes locally; the heavier regression harness defers to CI.

Change checklist

• I have performed a self-review of my own code,
• I have made corresponding changes to the documentation. All new code adheres to the team's documentation standards,
• I have added tests that prove my fix is effective or that my feature works (if possible),
• I have made sure the CHANGELOG is up-to-date. All user-facing changes should be reflected in this document.

Outside contributions

• I have read and agree to the CONTRIBUTING document.
• I have read and agree to the AI Policy document. I understand that failure to comply with the guidelines will lead to rejection of the pull request.

AI Usage Disclosure

This PR was prepared with assistance from Claude Code (Anthropic). Extent:

• Issue analysis, implementation, self-review, and PR drafting were AI-assisted.
• All changes were compiled and tested locally by me before pushing — cargo fmt --all -- --check, cargo clippy --all-targets --no-deps -- -D warnings (with FOREST_F3_SIDECAR_FFI_BUILD_OPT_OUT=1), cargo check --lib --tests, and cargo test --lib rpc_regression_tests_print_uncovered (1/1 pass) all green.
• I have reviewed every diff line and take full responsibility for correctness.

Summary by CodeRabbit

• Tests
  • Added explicit tracking and draining of background message-pool tasks used during snapshot tests.
  • Background tasks are now aborted promptly after setup and then awaited so their errors or panics are surfaced.
  • Improved logging for task errors and panics to make test failures and diagnostics clearer.

[CLAassistant]

All committers have signed the CLA.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: 4cef0c0e-8fde-477f-9134-23be3c604d0d

📥 Commits

Reviewing files that changed from the base of the PR and between e005443 and 6309c54.

📒 Files selected for processing (2)

• src/tool/subcommands/api_cmd/generate_test_snapshot.rs
• src/tool/subcommands/api_cmd/test_snapshot.rs

🚧 Files skipped from review as they are similar to previous changes (2)

• src/tool/subcommands/api_cmd/test_snapshot.rs
• src/tool/subcommands/api_cmd/generate_test_snapshot.rs

---

Walkthrough

A shared JoinSet<anyhow::Result<()>> is created and passed into MessagePool::new, immediately aborted, and then drained by a new drain_mpool_services async function that logs task errors and panics while ignoring cancellations.

Changes

JoinSet Error Handling in Test Utilities

Layer / File(s)	Summary
Create shared JoinSet and abort after MessagePool construction src/tool/subcommands/api_cmd/test_snapshot.rs, src/tool/subcommands/api_cmd/generate_test_snapshot.rs	Instantiates a shared JoinSet<anyhow::Result<()>>, passes it into MessagePool::new, and calls abort_all() immediately after construction; spawns a background task to drain the join set.
Drain JoinSet and log outcomes src/tool/subcommands/api_cmd/test_snapshot.rs	Adds pub(super) async fn drain_mpool_services(mut services: JoinSet<anyhow::Result<()>>) which loops on join_next(), logs warnings for task Err/panic results, and ignores cancelled joins.

Sequence Diagram

sequenceDiagram
  participant ctx
  participant MessagePool
  participant JoinSet
  participant drain_task

  ctx->>JoinSet: create JoinSet
  ctx->>MessagePool: MessagePool::new(&mut JoinSet)
  ctx->>JoinSet: abort_all()
  ctx->>drain_task: spawn drain_mpool_services(JoinSet)
  drain_task->>JoinSet: join_next() loop
  JoinSet-->>drain_task: Result (Ok / Err / Cancelled)
  drain_task->>drain_task: log warnings for Err and panic, ignore Cancelled

Loading

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

Suggested reviewers

• hanabi1224
• LesnyRumcajs

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately describes the main change: surfacing MessagePool service task errors in the api_cmd test context by implementing proper error handling.
Linked Issues check	✅ Passed	The PR fully implements the requirements from issue #6505: owns the JoinSet, adopts the canonical draining pattern with error logging, applies fixes to both target files, and prevents CI timeouts via abort_all() and detached draining.
Out of Scope Changes check	✅ Passed	All changes are directly related to resolving issue #6505. Modifications to both api_cmd test files focus solely on implementing proper JoinSet error handling and draining.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.

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

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

✨ Simplify code

• Create PR with simplified code

---

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.}

[LesnyRumcajs]

Seems to timeout the tests. Please first assert locally that it works.

[0xDevNinja]

Hey, thanks for catching that — local repro confirmed your concern. With just the detached drain_mpool_services spawn, the mpool background reactors stayed alive for the duration of each snapshot test. Running the full snapshot suite (370 tests) with --test-threads=8 had 5 tests stall on the parallel runtime under contention, exactly matching the codecov pattern from #6493.

Pushed a follow-up: services.abort_all() right after MessagePool::new(...) in both test_snapshot::ctx and generate_test_snapshot::ctx. The mpool services were never meant to run in this snapshot ctx (the inherited &mut JoinSet::new() pattern was a temporary that dropped immediately — same end state). The detached drain still polls the aborted JoinSet so any pre-abort spawn error or panic is surfaced via tracing::warn! instead of being silently dropped.

Locally:

$ FOREST_TEST_SKIP_PROOF_PARAM_CHECK=1 cargo test --lib --features cargo-test cargo_test_rpc_snapshot_test_filecoin_ -- --test-threads=8
test result: ok. 370 passed; 0 failed; 0 ignored; 0 measured; 2045 filtered out; finished in 31.87s

No over 60 seconds warnings, no panics. The single-commit branch is rebased on main and CI should pick it up shortly — happy to undraft once you take a look.