nautechsystems/nautilus_trader · error

Failed to commit order client origins: {e}

Error message

Failed to commit order client origins: {e}

What it means

At the end of index_order_clients the database transaction is committed; any sqlx commit failure is wrapped as 'Failed to commit order client origins'. The prior statements succeeded but finalizing them atomically failed — usually a lost connection or serialization/lock conflict at commit time. All changes roll back, so callers can safely retry the whole operation.

Source

Thrown at crates/infrastructure/src/sql/queries.rs:1504

                WHERE client_order_id = $1
                  AND (client_id IS NULL OR client_id = $2)
            "#,
            )
            .bind(client_order_id.to_string())
            .bind(client_id.to_string())
            .execute(&mut *transaction)
            .await
            .map_err(|e| anyhow::anyhow!("Failed to index order client origin: {e}"))?;

            if result.rows_affected() == 0 {
                anyhow::bail!("No persisted order events found for {client_order_id}");
            }
        }

        transaction
            .commit()
            .await
            .map_err(|e| anyhow::anyhow!("Failed to commit order client origins: {e}"))
    }

    /// Inserts or updates an order ID to position ID index entry via the provided `pool`.
    ///
    /// # Errors
    ///
    /// Returns an error if the SQL INSERT or UPDATE operation fails.
    pub async fn index_order_position(
        pool: &PgPool,
        client_order_id: ClientOrderId,
        position_id: PositionId,
    ) -> anyhow::Result<()> {
        sqlx::query(
            r#"
            INSERT INTO "order_position_index" (
                client_order_id, position_id, created_at, updated_at
            ) VALUES (
                $1, $2, CURRENT_TIMESTAMP, CURRENT_TIMESTAMP

View on GitHub (pinned to 18893faf8b)

Solutions

1. Simply retry index_order_clients — the rolled-back transaction leaves no partial state.
2. Check DB server logs for the commit failure cause (connection reset, serialization failure).
3. Keep the transaction short; batch large rebuilds into smaller transactions if timeouts occur.
4. Ensure pool timeouts and keepalives are configured for long operations.

Example fix

// before
queries::postgres::index_order_clients(&mut tx, &client_order_id, &client_id).await?;
// after
for attempt in 0..3 {
    match try_index_clients(&pool, &client_order_id, &client_id).await {
        Ok(_) => break,
        Err(e) if attempt < 2 && is_transient(&e) => tokio::time::sleep(backoff(attempt)).await,
        Err(e) => return Err(e),
    }
}

let healthy = pool.acquire().await.is_ok();
anyhow::ensure!(healthy, "database pool unhealthy before index commit");

for attempt in 0..3 {
    match index_clients_in_tx(&pool, &client_order_id, &client_id).await {
        Ok(_) => break,
        Err(e) if attempt < 2 && e.to_string().contains("commit") => {
            tokio::time::sleep(std::time::Duration::from_millis(250 * (attempt + 1))).await;
        }
        Err(e) => return Err(e),
    }
}

When it happens

Trigger: Calling index_order_clients when the connection drops between the UPDATEs and commit, the DB is shutting down, or a constraint/concurrency conflict surfaces only at COMMIT.

Common situations: Long-running index rebuild against a remote Postgres with an idle-timeout killing the connection; DB restart; network partition between nautilus and the cache DB.

Related errors

• Failed to commit verified nonce assignment: {e}
• Failed to commit transaction: {e}
• Failed to validate order client origin: {e}
• Unknown execution event marker {event}
• Failed to start verified action evidence: {e}

AI-assisted analysis of nautechsystems/nautilus_trader@18893faf8b (2026-09-08). Data as JSON: /api/errors/c331881866a3bcad. Report an issue: GitHub.