Monitoring processes in queues

You can monitor the handling processes for messages in queues using monitoring tools in the management console. These tools display diagnostic information as charts.

To view this information:

Sending messagesSending messages

Sent messages stats contains charts with information about sending messages:

• Sent messages, count: Number of messages sent to the queue per unit of time.

• Sent messages, size: The total size of messages sent to the queue per unit of time (in bytes).

• SendMessage duration: The time it takes to execute a SendMessage request.

  Displayed as percentiles. The lines on the chart correspond to the percentage of requests with the fastest execution time: 50%, 90%, 95%, and 99%.

• SendMessageBatch duration: The time it takes to execute a SendMessageBatch request.

  Displayed as percentiles. The lines on the chart correspond to the percentage of requests with the fastest execution time: 50%, 90%, 95%, and 99%.

• SendMessage errors: Number of errors when executing the SendMessage and SendMessageBatch requests.

Receiving messagesReceiving messages

Received messages & processing stats contains charts with information about receiving and processing messages:

• Received messages, size: The total size of messages received from the queue per unit of time (in bytes).

• Received messages, count: Number of messages received from the queue per unit of time.

• Messages in flight, count: Number of messages received by recipients but not deleted from the queue (that are still within the visibility timeout window).

• ReceiveMessages requests: Total number of ReceiveMessage requests and number of requests with empty responses.

  An empty response to a request indicates that there were no messages available for reading within the specified waiting time (the WaitTimeSeconds attribute).

• ReceiveMessages duration: The time it takes to execute a ReceiveMessage request.

  Displayed as percentiles. The lines on the chart correspond to the percentage of requests with the fastest execution time: 50%, 90%, 95%, and 99%.

• ReceiveMessage errors: Number of errors when executing the ReceiveMessage requests.

• Message processing on client duration: Message processing time spent by the recipient.

  Calculated as the difference between the time when the message was received using ReceiveMessage and when it was deleted using DeleteMessage or DeleteMessageBatch.
  Displayed as percentiles. The lines on the chart correspond to the percentage of messages with the fastest handling time: 50%, 90%, 95%, and 99%.

• Receive attempts per message, count: Number of registered attempts to receive a message.

  The value increases if a message was received by the recipient but returned to the queue after the visibility timeout expired. The lines on the chart correspond to the number of receive attempts:

  • 0-1
  • 2
  • 3-5
  • More than 5

Deleting messagesDeleting messages

Delete messages stats contains charts with information about deleting messages:

• Deleted messages, count: Number of messages deleted from the queue in a unit of time using the DeleteMessage or DeleteMessageBatch method.

• DeleteMessage duration: Time it takes to execute a DeleteMessage request.

  Displayed as percentiles. The lines on the chart correspond to the percentage of requests with the fastest execution time: 50%, 90%, 95%, and 99%.

• DeleteMessageBatch duration: The time it takes to execute a DeleteMessageBatch request.

  Displayed as percentiles. The lines on the chart correspond to the percentage of requests with the fastest execution time: 50%, 90%, 95%, and 99%.

• DeleteMessage errors: Number of errors when executing the DeleteMessage and DeleteMessageBatch requests.

• Purged messages, count: Number of messages purged from the queue in a unit of time using the PurgeQueue method.

Overall queue metricsOverall queue metrics

Overall queue stats contains charts with general information about processes in queues:

• Messages in queue, count (metric name: queue.messages.stored_count): Number of messages currently enqueued. Measured in pcs, number of.

• Age of oldest message in queue (metric name: queue.messages.oldest_age_milliseconds): Retention time of the oldest message in the queue. The metric name is obsolete and its milliseconds part does not reflect the current units of measure. Currently, the metric is measured in seconds (s).

• Message reside duration (metric names: p99 for the 99th percentile, p95 for the 95th percentile, p90 for the 90th percentile, p50 for the 50th percentile): Enqueued message processing time in milliseconds (ms).

  Calculated as the difference between the time when the message was received using ReceiveMessage and when it was deleted using DeleteMessage or DeleteMessageBatch.
  Displayed as percentiles. The lines on the chart correspond to the percentage of messages with the fastest handling time: 50%, 90%, 95%, and 99%.

See alsoSee also

Yandex Monitoring metric reference