overdrive

Redrive messages from a dead-letter queue (DLQ) back onto an SQS queue.

• Usage
• How it works
• Requirements

Usage

$ overdrive -h
Usage: overdrive [options]

Redrive dead-letter messages back into an SQS queue.

Options:
  -V, --version          output the version number
  -q, --queue-url <url>  SQS queue url to redrive (required)
  -t, --timeout <ms>     time to wait for DLQ to empty (default: 60000)
  -h, --help             output usage information

A common way to run commands like this is to include it as an npm script in your service's package.json.

{
  "scripts": {
    "sqs:redrive": "overdrive --queue-url $MY_SQS_QUEUE_URL"
  }
}

Then execute it with npm sqs:redrive. You should see animated output similar to the following:

Redriving 295 messages...
progress: ⣀ [========================] 100%
Done.

How it works

1. When you specify the main queue, overdrive will first query the queue attributes to parse the redrive-policy and obtain the DLQ url.

2. Next it will drain the DLQ and push all of the dead-letter messages back onto the original queue.

3. It ensures both the MessageBody and MessageAttributes are resent.

Requirements

When you do any work with SQS queues, you've got to have your AWS_REGION on hand, and that still applies here.

In addition, you'll need to get your IAM statements in order. Specific requirements will vary on your use-case, but assuming you've got one queue with a DLQ and both are encrypted with KMS, then your IAM statements should look something like this:

[{
  "Effect": "Allow",
  "Action": [
    "kms:Decrypt",
    "kms:GenerateDataKey"
  ],
  "Resource": "${kms_key.arn}"
}, {
  "Effect": "Allow",
  "Action": [
    "sqs:DeleteMessage",
    "sqs:GetQueueAttributes",
    "sqs:ReceiveMessage",
    "sqs:SendMessage"
  ],
  "Resource": "${queue.arn}"
}, {
  "Effect": "Allow",
  "Action": [
    "sqs:DeleteMessage",
    "sqs:GetQueueAttributes",
    "sqs:GetQueueUrl",
    "sqs:ReceiveMessage"
  ],
  "Resource": "${dead_letter_queue.arn}"
}]