Understanding Prometheus Matchers with Real Examples

Prometheus stores data in time series, and each time series has a metric name and labels (like handler, method, code, etc.). To filter data in Prometheus, we use matchers.

 

Let’s explore the most common matchers with simple examples.

 

1. Equality Matcher (=)

This checks if a label is exactly equal to the value (case sensitive).

 

Example

promhttp_metric_handler_requests_total{instance="localhost:9100"}

 

This query return the total number of requests handled by the instance "localhost:9100".

 

2. Negative Equality Matcher (!=)

This selects all time series except those where the label equals the value.

 

Example

promhttp_metric_handler_requests_total{code != "200"}

 

This query return the total number of requests for all handlers where code is not equal to 200.

 

3. Regular Expression Matcher (=~)

This allows pattern matching using regex. It's powerful when you want to match similar strings.

prometheus_http_requests_total{handler=~"/api/v1/status.+"}

 

Above query return all requests where the handler starts with /api/v1/status followed by anything else (like /api/v1/status/health, /api/v1/status/uptime, etc.). 

 

4. Negative Regex Matcher (!~)

This excludes label values that match the regex.

 

prometheus_http_requests_total{handler!~"/api/v1/status.+"}

 

Above query return all requests where the handler is not starting with /api/v1/status

 

In summary, Match selectors are a powerful way to filter data in Prometheus. Start with simple equality, then move to regex as your queries get more advanced.  

 

Previous                                                    Next                                                    Home