import SystemTableCloud from '@site/docs/_snippets/_system_table_cloud.md';

histogram_metrics {#histogram_metrics}

This table contains histogram metrics that can be calculated instantly and exported in the Prometheus format. It is always up to date. Replaces the deprecated system.latency_log.

Columns:

Example

You can use a query like this to export all the histogram metrics in the Prometheus format.

SELECT
  metric AS name,
  toFloat64(value) AS value,
  description AS help,
  labels,
  'histogram' AS type
FROM system.histogram_metrics
FORMAT Prometheus

Metric descriptions {#metric_descriptions}

keeper_response_time_ms_bucket {#keeper_response_time_ms_bucket}

The response time of Keeper, in milliseconds.

keeper_client_queue_duration_milliseconds_bucket {#keeper_client_queue_duration_milliseconds_bucket}

Time requests spend waiting to be enqueued and waiting in the queue before being processed by the Keeper client, in milliseconds.

keeper_receive_request_time_milliseconds_bucket {#keeper_receive_request_time_milliseconds_bucket}

Time to receive and parse a request from the client in the Keeper TCP handler, in milliseconds.

keeper_dispatcher_requests_queue_time_milliseconds_bucket {#keeper_dispatcher_requests_queue_time_milliseconds_bucket}

Time a request spends in the Keeper dispatcher requests queue, in milliseconds.

keeper_write_pre_commit_time_milliseconds_bucket {#keeper_write_pre_commit_time_milliseconds_bucket}

Time to preprocess a write request before Raft commit, in milliseconds.

keeper_write_commit_time_milliseconds_bucket {#keeper_write_commit_time_milliseconds_bucket}

Time to process a write request after Raft commit, in milliseconds.

keeper_dispatcher_responses_queue_time_milliseconds_bucket {#keeper_dispatcher_responses_queue_time_milliseconds_bucket}

Time a response spends in the Keeper dispatcher responses queue, in milliseconds.

keeper_send_response_time_milliseconds_bucket {#keeper_send_response_time_milliseconds_bucket}

Time to send a response to the client in the Keeper TCP handler (includes queueing and writing to socket), in milliseconds.

keeper_read_wait_for_write_time_milliseconds_bucket {#keeper_read_wait_for_write_time_milliseconds_bucket}

Time a read request waits for the write request it depends on to complete, in milliseconds.

keeper_read_process_time_milliseconds_bucket {#keeper_read_process_time_milliseconds_bucket}

Time to process a read request in Keeper, in milliseconds.

keeper_batch_size_elements_bucket {#keeper_batch_size_elements_bucket}

Batch size sent to Raft, in elements.

keeper_batch_size_bytes_bucket {#keeper_batch_size_bytes_bucket}

Batch size sent to Raft, in bytes.

See Also

• system.asynchronous_metrics — Contains periodically calculated metrics.
• system.events — Contains a number of events that occurred.
• system.metric_log — Contains a history of metrics values from tables system.metrics and system.events.
• Monitoring — Base concepts of ClickHouse monitoring.