All 15 Error Handling questions
More in Error Handling
What is the difference between throwing an error synchronously and calling a callback with an error as its first argument in Node.js? How would you create and use custom Error subclasses in a Node.js application to represent different categories of application-specific errors? Why doesn't Express automatically catch errors thrown inside an async route handler in versions prior to Express 5, and how do you work around this? How do you correctly propagate and handle errors from a Node.js stream, given that stream errors aren't caught by a surrounding try/catch? What is the risk of exposing a raw error's stack trace in an API response, and how do you prevent it while still logging full details server-side? What is the difference between a validation error and a business-logic error, and how might a Node.js application handle each differently? How would you implement a global 'not found' (404) handler in Express for any route that doesn't match a defined endpoint? What is a centralized error-handling middleware pattern in Express, and what advantages does it provide over handling errors in each individual route? What is the difference between a 4xx and 5xx HTTP status code from an error-handling perspective, and how should a Node.js API decide which to return? What is error wrapping (or error chaining), and how does the Error class's 'cause' property support it in modern Node.js?
Advertisement
Node.js › Error Handling

What is the try/catch/finally pattern's 'finally' block used for, and how does it behave with async/await?

Intermediate

Answer

The 'finally' block runs regardless of whether the try block succeeded or the catch block handled an error, making it the natural place for cleanup logic (closing a database connection, releasing a lock, stopping a loading spinner) that must happen in every case -- with async/await, a finally block correctly waits for the try/catch's async operations to settle before running, and importantly, an async operation inside finally itself is also properly awaited if the finally block itself is async.
async function processWithConnection(id) {
  const connection = await pool.connect();
  try {
    return await connection.query('SELECT ...', [id]);
  } catch (err) {
    logger.error(err);
    throw err;
  } finally {
    connection.release(); // always runs, whether the query succeeded or failed
  }
}
Real-world example A database query wrapper always releases its acquired connection back to the pool inside a finally block, guaranteeing the connection is returned regardless of whether the query succeeds, throws a validation error, or throws an unexpected database error.

Common follow-ups: What happens if the finally block itself throws an error -- does it override an error from the try or catch block?;Why is placing cleanup logic in finally more reliable than duplicating it at the end of both the try and catch blocks?

Databases & ORMs (MongoDB/Mongoose SQL/Sequelize);Async Patterns
All 15 Error Handling questions