What causes "connect to test DB" in block/buzz?

Start the test infrastructure: `just test` (or docker compose up the Postgres service) before running cargo test.

connect to test DB

Error message

connect to test DB

What it means

Panic in the shared `setup_db` test helper: `PgPool::connect(&database_url).await.expect("connect to test DB")`. It fires when the test suite cannot establish a Postgres connection using TEST_DATABASE_URL (or the built-in TEST_DB_URL default). Every test in this module calls setup_db, so a connection failure fails them all with this message.

Source

Thrown at crates/buzz-db/src/store/community.rs:567

mod tests {
    //! Pin the load-bearing contract for `Db::communities_of_channels`:
    //! a channel id that does NOT exist MUST be absent from the result
    //! map, never mapped to a default. The relay-side read-row emitter
    //! relies on this — a missing entry triggers `MissingLookup →
    //! ImplBug{row_community_lookup_missing} → CoverageBreach`. If this
    //! helper ever started returning a default/zero entry for unknown
    //! channels, that fail-closed chain would go blind.
    use super::*;
    use sqlx::PgPool;

    const TEST_DB_URL: &str = "postgres://buzz:buzz_dev@localhost:5432/buzz"; // sadscan:disable np.postgres.1

    async fn setup_db() -> Db {
        let database_url =
            std::env::var("TEST_DATABASE_URL").unwrap_or_else(|_| TEST_DB_URL.into());
        let pool = PgPool::connect(&database_url)
            .await
            .expect("connect to test DB");
        Db::from_pool(pool)
    }

    async fn make_community(pool: &PgPool) -> Uuid {
        let id = Uuid::new_v4();
        let host = format!("communities-of-channels-{}.example", id.simple());
        sqlx::query("INSERT INTO communities (id, host) VALUES ($1, $2)")
            .bind(id)
            .bind(host)
            .execute(pool)
            .await
            .expect("insert community");
        id
    }

    async fn insert_channel(pool: &PgPool, community_id: Uuid, channel_id: Uuid) {
        let creator: Vec<u8> = vec![0u8; 32];
        sqlx::query(

View on GitHub (pinned to eed74bde2f)

Solutions

Start the test infrastructure: `just test` (or docker compose up the Postgres service) before running cargo test.
Export TEST_DATABASE_URL to a reachable Postgres, e.g. postgres://postgres:postgres@localhost:5432/buzz_test.
Verify credentials and database name; create the test database if missing.
If connections are exhausted, close idle pools in tests (drop(pool)) or raise max_connections.

Example fix

// before
let database_url = std::env::var("TEST_DATABASE_URL").unwrap_or_else(|_| TEST_DB_URL.into());
let pool = PgPool::connect(&database_url).await.expect("connect to test DB");
// after
let database_url = std::env::var("TEST_DATABASE_URL").unwrap_or_else(|_| TEST_DB_URL.into());
let pool = PgPool::connect(&database_url).await
    .unwrap_or_else(|e| panic!("connect to test DB at {database_url}: {e}"));

Defensive patterns

Strategy: validation

Validate before calling

fn require_test_database_url() -> String {
    std::env::var("TEST_DATABASE_URL")
        .unwrap_or_else(|_| TEST_DB_URL.into())
}
// pre-flight check before the suite:
let url = require_test_database_url();
PgPool::connect(&url).await.expect("TEST_DATABASE_URL unreachable — start Postgres (just test / docker compose up)");

Try / catch

let pool = PgPool::connect(&database_url).await
    .unwrap_or_else(|e| panic!("connect to test DB ({database_url}): {e} — is Postgres running and TEST_DATABASE_URL set?"));

Prevention

Always run integration tests through `just test`, which provisions Postgres.
Set TEST_DATABASE_URL in .env and keep it consistent across worktrees.
Add a pre-flight connectivity check before running long suites.
In CI, verify the Postgres service container is healthy before cargo test.

When it happens

Trigger: TEST_DATABASE_URL unset and the default URL unreachable; Postgres container not started; wrong host/port/user/password/database name; Postgres up but max_connections exhausted; DNS failure for a non-localhost host.

Common situations: Running `cargo test` directly instead of `just test` (which starts the Docker Postgres), switching between worktrees with different .env values, CI missing the Postgres service container, or a stale TEST_DATABASE_URL pointing at a torn-down scratch instance.

Related errors

AI-assisted analysis of block/buzz@eed74bde2f (2026-08-30). Data as JSON: /api/errors/79c07c58eb43c516. Report an issue: GitHub.