Retry, timeout and errors

Retry

const results = await batchProcess(
  unreliableOperations,
  async (operation) => await performOperation(operation),
  {
    retry: {
      maxAttempts: 3,
      backoff: 'exponential', // or 'linear' (default)
      initialDelay: 100,      // default: 100ms
      maxDelay: 30000         // default: 30000ms
    }
  }
)

• maxAttempts is the total number of attempts, including the first one. Values below 1 are treated as 1 and non-integers are floored.
• linear waits initialDelay before every retry.
• exponential waits initialDelay * 2^(attempt - 1), capped at maxDelay.

Timeout

timeout limits each attempt of each task. A task that exceeds it rejects with Error('Task timed out after <timeout>ms'), which then goes through the retry logic like any other error.

try {
  await batchProcess(tasks, async (task) => await performTask(task), {
    timeout: 5000
  })
} catch (error) {
  console.error('Task timed out:', error.message)
}

Caution

The timeout only stops waiting for the task. The underlying promise is not cancelled, so pass your own AbortSignal to the task if it must actually stop.

Continue on error

By default the first error that survives all retries rejects the whole batchProcess call. With continueOnError: true, the error is placed in the results at the item’s position instead.

const results = await batchProcess<Item, Result | Error>(
  items,
  async (item) => await processItem(item),
  { continueOnError: true }
)

results.flat().forEach((result, index) => {
  if (result instanceof Error) {
    console.error(`Item ${index} failed:`, result.message)
  }
})

Declare the result type as Result | Error so TypeScript reminds you to handle the error case.