Defining recording rules

Configuring rules

Prometheus supports two types of rules which may be configured and then evaluated at regular intervals: recording rules and alerting rules. To include rules in Prometheus, create a file containing the necessary rule statements and have Prometheus load the file via the rule_files field in the Prometheus configuration. Rule files use YAML.

The rule files can be reloaded at runtime by sending SIGHUP to the Prometheus process. The changes are only applied if all rule files are well-formatted.

Note about native histograms (experimental feature): Native histogram are always recorded as gauge histograms (for now). Most cases will create gauge histograms naturally, e.g. after rate().

Syntax-checking rules

To quickly check whether a rule file is syntactically correct without starting a Prometheus server, you can use Prometheus’s promtool command-line utility tool:

  1. promtool check rules /path/to/example.rules.yml

The promtool binary is part of the prometheus archive offered on the project’s download page.

When the file is syntactically valid, the checker prints a textual representation of the parsed rules to standard output and then exits with a 0 return status.

If there are any syntax errors or invalid input arguments, it prints an error message to standard error and exits with a 1 return status.

Recording rules

Recording rules allow you to precompute frequently needed or computationally expensive expressions and save their result as a new set of time series. Querying the precomputed result will then often be much faster than executing the original expression every time it is needed. This is especially useful for dashboards, which need to query the same expression repeatedly every time they refresh.

Recording and alerting rules exist in a rule group. Rules within a group are run sequentially at a regular interval, with the same evaluation time. The names of recording rules must be valid metric names. The names of alerting rules must be valid label values.

The syntax of a rule file is:

  1. groups:
  2.   [ - <rule_group> ]

A simple example rules file would be:

  1. groups:
  2.   - name: example
  3.     rules:
  4.     - record: code:prometheus_http_requests_total:sum
  5.       expr: sum by (code) (prometheus_http_requests_total)

<rule_group>

  1. # The name of the group. Must be unique within a file.
  2. name: <string>
  4. # How often rules in the group are evaluated.
  5. [ interval: <duration> | default = global.evaluation_interval ]
  7. # Limit the number of alerts an alerting rule and series a recording
  8. # rule can produce. 0 is no limit.
  9. [ limit: <int> | default = 0 ]
  11. rules:
  12.   [ - <rule> ... ]

<rule>

The syntax for recording rules is:

  1. # The name of the time series to output to. Must be a valid metric name.
  2. record: <string>
  4. # The PromQL expression to evaluate. Every evaluation cycle this is
  5. # evaluated at the current time, and the result recorded as a new set of
  6. # time series with the metric name as given by 'record'.
  7. expr: <string>
  9. # Labels to add or overwrite before storing the result.
  10. labels:
  11.   [ <labelname>: <labelvalue> ]

The syntax for alerting rules is:

  1. # The name of the alert. Must be a valid label value.
  2. alert: <string>
  4. # The PromQL expression to evaluate. Every evaluation cycle this is
  5. # evaluated at the current time, and all resultant time series become
  6. # pending/firing alerts.
  7. expr: <string>
  9. # Alerts are considered firing once they have been returned for this long.
  10. # Alerts which have not yet fired for long enough are considered pending.
  11. [ for: <duration> | default = 0s ]
  13. # How long an alert will continue firing after the condition that triggered it
  14. # has cleared.
  15. [ keep_firing_for: <duration> | default = 0s ]
  17. # Labels to add or overwrite for each alert.
  18. labels:
  19.   [ <labelname>: <tmpl_string> ]
  21. # Annotations to add to each alert.
  22. annotations:
  23.   [ <labelname>: <tmpl_string> ]

See also the best practices for naming metrics created by recording rules.

Limiting alerts and series

A limit for alerts produced by alerting rules and series produced recording rules can be configured per-group. When the limit is exceeded, all series produced by the rule are discarded, and if it’s an alerting rule, all alerts for the rule, active, pending, or inactive, are cleared as well. The event will be recorded as an error in the evaluation, and as such no stale markers are written.

Failed rule evaluations due to slow evaluation

If a rule group hasn’t finished evaluating before its next evaluation is supposed to start (as defined by the evaluation_interval), the next evaluation will be skipped. Subsequent evaluations of the rule group will continue to be skipped until the initial evaluation either completes or times out. When this happens, there will be a gap in the metric produced by the recording rule. The rule_group_iterations_missed_total metric will be incremented for each missed iteration of the rule group.

This documentation is open-source. Please help improve it by filing issues or pull requests.