View a markdown version of this page

Monitor a queue - Amazon Kinesis Data Streams
Services or capabilities described in Amazon Web Services documentation might vary by Region. To see the differences applicable to the China Regions, see Getting Started with Amazon Web Services in China (PDF).

Monitor a queue

When you consume records from a stream with a queue, Kinesis Data Streams sends metrics for the queue to CloudWatch every minute in the Amazon/Kinesis namespace. Kinesis Data Streams publishes these metrics with the QueueId and QueueName dimensions.

Note

After you create a queue, Kinesis Data Streams can take up to three minutes to publish its first metrics for the queue. Kinesis Data Streams also stops publishing metrics for a queue that has had no visible records for about six hours, and publishes them again as soon as a consumer instance receives from the queue.

The following table describes the metrics that Kinesis Data Streams publishes for the records in a queue.

Metric Description
ReceivedQueueRecords

The number of records that queue consumers received from the queue over the specified time period.

Dimensions: QueueId, QueueName

Statistics: Sum

Units: Count

AcknowledgedQueueRecords

The number of records that queue consumers acknowledged over the specified time period.

Dimensions: QueueId, QueueName

Statistics: Sum

Units: Count

EmptyReceives

The number of receive calls that returned no records over the specified time period.

Dimensions: QueueId, QueueName

Statistics: Sum

Units: Count

VisibleQueueRecords

The approximate number of records in the queue that are visible and ready to be received. This metric reports up to 1,000,000 records.

Dimensions: QueueId, QueueName

Statistics: Average

Units: Count

InFlightQueueRecords

The approximate number of records in the queue that are in flight: a consumer instance has received each one, and has not acknowledged it, and the queue has not yet made it visible again.

Dimensions: QueueId, QueueName

Statistics: Average

Units: Count

The following table describes the metrics that Kinesis Data Streams publishes for the dead-letter records in a queue. We recommend that you create a CloudWatch alarm on VisibleDeadLetterQueueRecords to detect records that a queue consumer cannot process.

Metric Description
ReceivedDeadLetterQueueRecords

The number of dead-letter records that queue consumers received over the specified time period.

Dimensions: QueueId, QueueName

Statistics: Sum

Units: Count

AcknowledgedDeadLetterQueueRecords

The number of dead-letter records that queue consumers acknowledged over the specified time period.

Dimensions: QueueId, QueueName

Statistics: Sum

Units: Count

EmptyDeadLetterReceives

The number of dead-letter receive calls that returned no records over the specified time period.

Dimensions: QueueId, QueueName

Statistics: Sum

Units: Count

VisibleDeadLetterQueueRecords

The approximate number of dead-letter records that are visible and ready to be received. This metric reports up to 1,000,000 records.

Dimensions: QueueId, QueueName

Statistics: Average

Units: Count

RestoredDeadLetterQueueRecords

The number of dead-letter records that RestoreDeadLetterQueueRecords returned to the normal receive path over the specified time period. Kinesis Data Streams publishes this metric only after you have restored at least one dead-letter record in the queue.

Dimensions: QueueId, QueueName

Statistics: Sum

Units: Count

InFlightDeadLetterQueueRecords

The approximate number of dead-letter records that are in flight.

Dimensions: QueueId, QueueName

Statistics: Average

Units: Count