[loganj]

🤖

Summary

The relay's PostgreSQL CI job sometimes fails in the channel group-state tests with ensure community: Sqlx(PoolTimedOut). The tests themselves are fine. They connect to the test database through a test helper that waits only 100 ms for a connection, and on a busy CI runner the first connection can take longer than that. One example is group_state_events_carry_type_then_creator in this #8522 run: it failed after 0.57 s, while its other 1110 tests passed.

This PR gives these tests a pool with sqlx's normal acquire timeout. No product code changes.

Details

• test_state_with_database_url keeps its 100 ms timeout on purpose. Its other callers point at a closed port (127.0.0.1:1) to test fail-closed paths, and the short timeout makes them fail fast.
• group_state_postgres_tests::state() is the only caller that passes it the real test database. It now builds a normal lazy pool and passes it to test_state_with_database_pool. The same state() also serves create_with_members_postgres_tests, edit_metadata_postgres_tests, and system_channel_postgres_tests.

Related issue

N/A. I searched open PRs for PoolTimedOut and acquire_timeout and found none for this.

Testing

To reproduce a slow first connection, I ran the tests through a local TCP proxy that waits 250 ms before it connects to Postgres:

• Before this change: group_state_postgres_tests had 13 of 14 tests fail with PoolTimedOut.
• After this change: all 14 pass. create_with_members_postgres_tests (3), edit_metadata_postgres_tests (7), and system_channel_postgres_tests (9) also pass through the same proxy.
• Without the proxy, all 14 pass before and after.
• cargo clippy -p buzz-relay --all-targets -D warnings and just test-unit pass.

[github-actions]

🔐 Codex Security Review

Note: This is an automated, security-focused review generated by Codex.
Use it as a supplement to human review; false positives are possible.

Scope

• Exact PR diff: 3ce3a2e2df815924f2519e08d5626a8f8bd8587b...0e30c380bdd93672ce95a36b735993af5648abb2
• Model: gpt-6.1-sol

💡 Click "edited" above to see earlier reviews for this PR.

---

Review Summary

Overall Risk: NONE

No concrete findings in the authorized PR range. Changes are confined to test helpers, preserve the existing 100 ms timeout for failure-path tests, and use default pool options for group-state Postgres tests.

Findings

No concrete security, correctness, or reliability findings were identified.

Notes

• Review used read-only source and history inspection. Builds and tests were not executed, as requested.

---

_{Generated by Codex Security Review |
Requested by: @loganj |
Workflow run}

[jedwards27]

🤖 :bot: Jude’s code review agent review

Verdict: APPROVE
Reviewed: PR diff at exact head 0e30c380bdd93672ce95a36b735993af5648abb2 (block/buzz#8564), live base ref 3ce3a2e2df815924f2519e08d5626a8f8bd8587b; local merge-base f167c4c3e10469f4f75b5ecad83af50bb3d5de2c. Authenticated reviewer jedwards27; PR author loganj.

Risk: Low. This is a relay test-fixture-only change: no production path, schema, or user-visible behavior changes. The main tradeoff is that failed group-state Postgres fixture acquisition can now wait for SQLx’s normal finite acquire timeout instead of the previous 100 ms CI-flake window.

Behavior/contracts traced:

• group_state_postgres_tests::state() now passes test_support::database_url() into the new URL+pool-options helper and uses PgPoolOptions::new(), so the shared real-Postgres group-state fixture keeps SQLx’s default acquire behavior instead of the artificial 100 ms cap (crates/buzz-relay/src/handlers/group_state_postgres_tests.rs:15-24).
• The helper sets both Config.database_url and the lazy writer pool from the same database_url, fixing the earlier config/pool mismatch concern (crates/buzz-relay/src/state.rs:2984-2999).
• test_state_with_database_url(...) still applies the 100 ms acquire timeout for explicit failure/closed-port seams (crates/buzz-relay/src/state.rs:2970-2982). Scoped callers are closed-port/fail-closed tests.
• The PostgreSQL nextest wrapper isolates tests by exporting DATABASE_URL, TEST_DATABASE_URL, and BUZZ_TEST_DATABASE_URL to the same per-test database (scripts/postgres-test-wrapper.sh:81-90).

Findings: No blocking or non-blocking code defects found. The older CHANGES_REQUESTED concern about the real DB pool using a different URL than config is resolved by the current head.

Author action: None required. Optional only: refresh the PR body’s implementation detail that still describes the intermediate test_state_with_database_pool shape; current code uses test_state_with_database_url_and_pool_options.

Verification owner: No author-owned blocker. CI covers the full Postgres lane; reviewer/tooling owns the residual local delayed-TCP-proxy repro gap.

Validation at matching head:

• git rev-parse HEAD → 0e30c380bdd93672ce95a36b735993af5648abb2; git status --short clean.
• gh pr view 8564 --repo block/buzz --json headRefOid,baseRefOid,author → head 0e30c380bdd93672ce95a36b735993af5648abb2, base 3ce3a2e2df815924f2519e08d5626a8f8bd8587b, author loganj.
• gh pr checks 8564 --repo block/buzz --required --json name,state filtered to non-pass/non-skipped required checks → 0.
• git diff --check origin/main...HEAD — pass.
• scripts/test-postgres-test-discovery.sh — pass.
• cargo test -p buzz-relay --no-run — pass.
• cargo test -p buzz-relay --lib state::tests::bare_community_disconnect_fails_closed_when_the_fence_is_unavailable -- --exact — pass.

Manual/native evidence: Not applicable for a relay test-helper-only change. I did not rerun the ignored Postgres group-state tests or delayed TCP-proxy reproduction locally because no local Docker/Postgres service was available in this session; exact-head CI’s required Postgres gates are green.

Residual risk: If a real group-state test pool starvation occurs, failure can now take up to SQLx’s normal default acquire timeout rather than 100 ms. That is finite and diagnostic, and the closed-port failure tests keep the short timeout.