Skip to main content
Skip to main content
Edit this page

system.aggregated_zookeeper_log

This table contains aggregated statistics of ZooKeeper operations (e.g. number of operations, average latency, errors) grouped by (session_id, parent_path, operation, component, is_subrequest) and periodically flushed to disk.

Unlike system.zookeeper_log which logs every individual request and response, this table aggregates operations into groups, making it much more lightweight and therefore more suitable for production workloads.

Operations that are part of a Multi or MultiRead batch are tracked separately via the is_subrequest column. Subrequests have zero latency because the total latency is attributed to the enclosing Multi/MultiRead operation.

Columns:

  • hostname (LowCardinality(String)) — Hostname of the server.
  • event_date (Date) — Date the group was flushed.
  • event_time (DateTime) — Time the group was flushed.
  • session_id (Int64) — Session id.
  • parent_path (String) — Prefix of the path.
  • operation (Enum) — Type of ZooKeeper operation.
  • is_subrequest (UInt8) — Whether this operation was a subrequest inside a Multi or MultiRead operation.
  • count (UInt32) — Number of operations in the group.
  • errors (Map(Enum, UInt32)) — Errors in the group, mapping error code to count.
  • average_latency (Float64) — Average latency across all operations in the group, in microseconds. Subrequests have zero latency because the latency is attributed to the enclosing Multi or MultiRead operation.
  • component (LowCardinality(String)) — Component that caused the event.

See Also