histogram_metrics

Description

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.

Metric descriptions

Metric	Description
`keeper_response_time_ms_bucket`	The response time of Keeper, in milliseconds.
`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`	Time to receive and parse a request from the client in the Keeper TCP handler, in milliseconds.
`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`	Time to preprocess a write request before Raft commit, in milliseconds.
`keeper_write_commit_time_milliseconds_bucket`	Time to process a write request after Raft commit, in milliseconds.
`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`	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`	Time a read request waits for the write request it depends on to complete, in milliseconds.
`keeper_read_process_time_milliseconds_bucket`	Time to process a read request in Keeper, in milliseconds.
`keeper_batch_size_elements_bucket`	Batch size sent to Raft, in elements.
`keeper_batch_size_bytes_bucket`	Batch size sent to Raft, in bytes.
`filesystem_cache_evicted_segment_hits_bucket`	Distribution of cache-hit counts on file segments at the moment of their eviction, labelled by cache name.
`filesystem_cache_evicted_segment_size_bytes_bucket`	Distribution of byte sizes of evicted file segments, labelled by cache name.
`filesystem_cache_evicted_segment_hits_by_user_bucket`	Distribution of cache-hit counts on evicted file segments, labelled by cache name and user id.
`filesystem_cache_evicted_segment_size_bytes_by_user_bucket`	Distribution of byte sizes of evicted file segments, labelled by cache name and user id.

Columns

metric (String) — Metric name.
value (Float64) — Metric value.
description (String) — Metric description.
labels (Map(String, String)) — Metric labels.

Aliases:

name — Alias for metric.

Examples

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