InfluxDB Enterprise query limits#

This guide describes the InfluxDB Enterprise query limits configuration used in Sasquatch.

InfluxDB Enterprise supports several mechanisms to control resource consumption by user queries. These controls are important to protect the database from excessive load caused by poorly shaped queries or unexpected query spikes.

The default query limits configuration used in Sasquatch is specified in the influxdb-enterprise/values.yaml file.

The following configuration options are set:

config:
  cluster:
      max-concurrent-queries: 50
      query-timeout: "180s"

      max-select-point: 50000000
      max-select-series: 200000
      max-select-buckets: 20000

      log-queries-after: "10s"
      log-timedout-queries: true
      termination-query-log: true

  • max-concurrent-queries limits the number of concurrent queries that can be executed by the cluster. This is not a rate limit, but a limit on simultaneous queries. It can be monitored from the queriesActive and queryExecutorBusy InfluxDB internal metrics.

  • query-timeout specifies the maximum duration a query can run before being automatically terminated by the database.

These limits kill or reject excessively expensive individual queries before they destabilize the cluster:

  • max-select-point Maximum total points a single SELECT can process. Example: limit SELECT over very long time ranges.

  • max-select-series Maximum series a single SELECT can process. Example: limit SELECT * on high-cardinality datasets.

  • max-select-buckets Maximum number of GROUP BY time() buckets. Example: limit GROUP BY time(1s) over weeks of data.

In addition, the following logging options are configured to help monitor and debug long-running queries:

  • log-queries-after specifies the duration after which a running query is logged to the InfluxDB logs.

  • log-timedout-queries enables logging of queries that time out.

  • termination-query-log enables logging of queries that are terminated due to resource limits.