High-volume Amazon SQS Poller and single-queue client for Node.js 4 and up
const poller = new Squiss({
queueName: 'my-sqs-queue',
bodyFormat: 'json',
unwrapSns: true,
maxInFlight: 500
});
poller.start();
poller.on('message', (msg) => {
console.log('%s says: %s', msg.body.name, msg.body.message);
msg.del();
});Squiss aims to process as many messages simultaneously as possible. Set the maxInFlight option to the number of messages your app can handle at one time without choking, and Squiss will attempt to keep that many messages flowing through your app, grabbing more as you mark each message as handled or ready for deletion. If the queue is empty, Squiss will maintain an open connection to SQS, waiting for any messages that appear in real time. By default, anyway. You can configure it to poll however you want. Squiss don't care.
Don't be scared of new -- you need to create a new Squiss instance for every queue you want to poll. Squiss is an EventEmitter, so don't forget to call squiss.on('message', (msg) => msg.del()) at the very least.
Use the following options to point Squiss at the right queue:
- opts.awsConfig An object mapping to pass to the SQS constructor, configuring the aws-sdk library. This is commonly used to set the AWS region, endpoint, or the user credentials. See the docs on configuring the aws-sdk for details.
- opts.queueName The name of the queue to be polled. Used only if opts.queueUrl is not specified, but Squiss prefers just the name.
- opts.queueUrl The URL of the queue to be polled. If not specified, opts.queueName is required.
- opts.accountNumber If a queueName is specified, the accountNumber of the queue owner can optionally be specified to access a queue in a different AWS account.
- opts.correctQueueUrl Default false. Changes the protocol, host, and port of the queue URL to match the configured SQS endpoint (see opts.awsConfig), applicable only if opts.queueName is specified. This can be useful for testing against a local SQS service, such as ElasticMQ.
Squiss's defaults are great out of the box for most use cases, but you can use the below to fine-tune your Squiss experience:
- opts.SQS Default AWS.SQS An instance of the official SQS Client, or an SQS constructor function to use rather than the default one provided by AWS.SQS
- opts.activePollIntervalMs Default 0. The number of milliseconds to wait between requesting batches of messages when the queue is not empty, and the maxInFlight cap has not been hit. For most use cases, it's better to leave this at 0 and let Squiss manage the active polling frequency according to maxInFlight.
- opts.advancedCallMs Default 5000. If
opts.autoExtendTimeoutis used, this is the number of milliseconds that Squiss will make the call to extend the VisibilityTimeout of the message before the message is set to expire. - opts.autoExtendTimeout Default false. If true, Squiss will automatically extend each message's VisibilityTimeout in the SQS queue until it's handled (by keeping, deleting, or releasing it). It will place the API call to extend the timeout
opts.advancedCallMsmilliseconds in advance of the expiration, and will extend it by the number of seconds specified inopts.visibilityTimeoutSecs. If that's not specified, the VisibilityTimeout setting on the queue itself will be used. - opts.bodyFormat Default "plain". The format of the incoming message. Set to "json" to automatically call
JSON.parse()on each incoming message. - opts.deleteBatchSize Default 10. The number of messages to delete at one time. Squiss will trigger a batch delete when this limit is reached, or when deleteWaitMs milliseconds have passed since the first queued delete in the batch; whichever comes first. Set to 1 to make all deletes immediate. Maximum 10.
- opts.deleteWaitMs Default 2000. The number of milliseconds to wait after the first queued message deletion before deleting the message(s) from SQS.
- opts.idlePollIntervalMs Default 0. The number of milliseconds to wait before requesting a batch of messages when the queue was empty on the prior request.
- opts.maxInFlight Default 100. The number of messages to keep "in-flight", or processing simultaneously. When this cap is reached, no more messages will be polled until currently in-flight messages are marked as deleted or handled. Setting this option to 0 will uncap your inFlight messages, pulling and delivering messages as long as there are messages to pull.
- opts.noExtensionsAfterSecs Default 43200. If
opts.autoExtendTimeoutis used, Squiss will stop auto-renewing a message's VisibilityTimeout when it reaches this age. Default is 12 hours, SQS's VisbilityTimeout maximum. - opts.pollRetryMs Default 2000. The number of milliseconds to wait before retrying when Squiss's call to retrieve messages from SQS fails.
- opts.receiveBatchSize Default 10. The number of messages to receive at one time. Maximum 10 or maxInFlight, whichever is lower.
- opts.receiveWaitTimeSecs Default 20. The number of seconds for which to hold open the SQS call to receive messages, when no message is currently available. It is recommended to set this high, as Squiss will re-open the receiveMessage HTTP request as soon as the last one ends. If this needs to be set low, consider setting activePollIntervalMs to space out calls to SQS. Maximum 20.
- opts.unwrapSns Default false. Set to
trueto denote that Squiss should treat each message as though it comes from a queue subscribed to an SNS endpoint, and automatically extract the message from the SNS metadata wrapper. - opts.visibilityTimeoutSecs Defaults to queue setting on read, or 30 seconds for createQueue. The amount of time, in seconds, that received messages will be unavailable to other pollers without being deleted.
Are you using Squiss to create your queue, as well? Squiss will use opts.receiveWaitTimeSecs and opts.visibilityTimeoutSecs above in the queue settings, but consider setting any of the following options to configure it further. Note that the defaults are the same as Amazon's own:
- opts.delaySecs Default 0. The number of milliseconds by which to delay the delivery of new messages into the queue by default.
- opts.maxMessageBytes Default 262144 (256KB). The maximum size of a single message, in bytes, that the queue can support.
- opts.messageRetentionSecs Default 345600 (4 days). The amount of time for which to retain messages in the queue until they expire, in seconds. Maximum is 1209600 (14 days).
- opts.queuePolicy If specified, will be set as the access policy of the queue when
createQueueis called. See the AWS Policy documentation for more information.
Creates the configured queue! This returns a promise that resolves with the new queue's URL when it's complete. Note that this can only be called if you set opts.queueName when instantiating Squiss.
Deletes a message, given the full Message object sent to the message event. It's much easier to call message.del(), but if you need to do it right from the Squiss instance, this is how. Note that the message probably won't be deleted immediately -- it'll be queued for a batch delete. See the constructor notes for how to configure the specifics of that.
Changes the visibility timeout of a message, given either the full Squiss Message object or the receipt handle string.
Deletes the configured queue, returning a promise that resolves on complete. Squiss lets you do this, even though it makes Squiss useless. Squiss is so selfless.
Returns a Promise that resolves with the URL of the configured queue, even if you only instantiated Squiss with a queueName. The correctQueueUrl setting applies to this result, if it was set.
Informs Squiss that you got a message that you're not planning on deleting, so that Squiss can decrement the number of "in-flight" messages. It's good practice to delete every message you process, but this can be useful in case of error. You can also call message.keep() on the message itself to invoke this.
Releases the given Message object back to the queue by setting its VisibilityTimeout to 0 and marking the message as handled internally. You can also call message.release() on the message itself to invoke this.
Sends an individual message to the configured queue, and returns a promise that resolves with AWS's official message metadata: an object containing MessageId, MD5OfMessageAttributes, and MD5OfMessageBody. Arguments:
- message. The message to push to the queue. If it's a string, great! If it's an Object, Squiss will call JSON.stringify on it.
- delay optional. The amount of time, in seconds, to wait before making the message available in the queue. If not specified, the queue's configured value will be used.
- attributes optional. An optional attributes mapping to associate with the message. For more information, see the official AWS documentation.
Sends an array of any number of messages to the configured SQS queue, breaking them down into appropriate batch requests executed in parallel (or as much as the default HTTP agent allows). It returns a promise that resolves with a response closely aligned to the official AWS SDK's sendMessageBatch, except the results from all batch requests are merged. Expect a result similar to:
{
Successful: [
{Id: string, MessageId: string, MD5OfMessageAttributes: string, MD5OfMessageBody: string}
],
Failed: [
{Id: string, SenderFault: boolean, Code: string, Message: string}
]
}The "Id" supplied in the response will be the index of the message in the original messages array, in string form. Arguments:
- messages. The array of messages to push to the queue. The messages should be either strings, or Objects that Squiss can pass to JSON.stringify.
- delay optional. The amount of time, in seconds, to wait before making the messages available in the queue. If not specified, the queue's configured value will be used.
- attributes optional. An optional attributes mapping to associate with each message. For more information, see the official AWS documentation.
Starts polling SQS for new messages. Each new message is handed off in the message event.
Hold on to your hats, this one stops the polling, aborting any in-progress request for new messages. If called with soft=true while there's an active request for new messages, the active request will not be aborted and the message event may still be fired up to opts.receiveWaitTimeSecs afterward.
The number of messages currently in-flight.
true if Squiss is actively polling SQS. If it's not polling, we made the genius design decision to have this set to false.
For your convenience, Squiss provides direct access to the AWS SDK's SQS object, which can be handy for setting up or tearing down tests. No need to thank Squiss. Squiss does this because Squiss cares.
Emitted when a message is confirmed as being successfully deleted from the queue. The handled and delQueued events will also be fired for deleted messages, but that will come earlier, when the delete function is initially called.
A delError is emitted when AWS reports that any of the deleted messages failed to actually delete. The
object handed to you in this event is the AWS failure object described in the SQS deleteMessageBatch documentation.
Emitted when a message is queued for deletion, even if delete queuing has been turned off.
Emitted when the last in-flight message has been handled, and there are no more messages currently in flight.
If any of the AWS API calls outrightly fail, error is emitted. If you don't have a listener on error, per Node.js's structure, the error will be treated as uncaught and will crash your app.
Emitted when Squiss asks SQS for a new batch of messages, and gets some (or one). Supplies the number of retrieved messages.
Emitted when a message is handled by any means: deleting, releasing, or calling keep() or handledMessage() on it.
Emitted when Squiss asks SQS for new messages, and doesn't get any.
Emitted after release() or releaseMessage has been called and the VisibilityTimeout of a message has successfully been changed to 0. The handled event will also be fired for released messages, but that will come earlier, when the release function is initially called.
Emitted when a message has had its timeout successfully extended by the autoExtendTimeout feature.
Emitted after a hard stop() if a request for new messages was already in progress.
Emitted if automatic VisibilityTimeout extensions are enabled, and Squiss attempts to extend a message that has either been deleted or otherwise expired.
Emitted when Squiss has hit the maxInFlight cap. At this point, Squiss won't retrieve any more messages until at least opts.receiveBatchSize in-flight messages have been deleted.
Emitted every time Squiss pulls a new message from the queue. The Squiss Message object handed back has the following methods and properties:
The body of the SQS message, unwrapped from the SNS metadata wrapper (if unwrapSns was specified in the constructor), and JSON-parsed (if bodyFormat: 'json' was specified in the constructor). Otherwise the body will just be a string.
The subject of the SNS message, if set. Exists only if unwrapSns was specified.
The full SNS topic ARN to which this message was posted. Exists only if unwrapSns was specified.
The name of the SNS topic to which this message was posted. Included for convenience only: this is the last segment of the topicArn. Exists only if unwrapSns was specified.
Deletes the message from SQS. Either this or message.keep() must be called on each message Squiss delivers in order to maintain an accurate inFlight count.
Instructs Squiss that you're not planning to delete a message, but it should no longer be considered "in-flight". Either this, message.del(), or message.release() must be called on each message Squiss delivers in order to maintain an accurate inFlight count.
Changes the visibility timeout of the message to 0.
Changes the visibility timeout of the message.
The raw, unprocessed SQS response object as delivered from the aws-sdk.
Squiss supports Node 4 LTS and higher. For 0.12 support, consider compiling with Babel or using Squiss version 0.x.
Squiss is Copyright (c) 2017 Tom Shawver, released under the ultra-permissive ISC license. See LICENSE.txt for details.
Squiss was originally created at TechnologyAdvice in Nashville, TN.