Query Language

Lakerunner uses query languages inspired by LogQL (opens in a new tab) (for logs) and PromQL (opens in a new tab) (for metrics). If you're familiar with either of these, you'll feel right at home — but note that Lakerunner supports a subset of their operators and syntax. We do not currently have full 1-to-1 compatibility with LogQL or PromQL.

Log Queries (LogQL-style)

Log queries retrieve and filter log entries. A query starts with a stream selector and can be followed by pipeline stages.

Stream Selectors

Stream selectors filter logs by label. They are enclosed in curly braces:

{label="value"}

Supported matching operators

Examples

{resource_service_name="api-server"}

{environment="prod", resource_service_name=~"api-.*"}

{environment!="dev"}

Line Filter Expressions

After a stream selector, you can apply line filters to match or exclude log content:

Examples

{resource_service_name="api-server"} |= "error"

{resource_service_name="api-server"} !~ "debug|trace"

{environment="prod"} |= "timeout" != "expected timeout"

Label Filter Expressions

Use label filters to refine results based on parsed or indexed labels:

{resource_service_name="api-server"} | log_level="ERROR"

{environment="prod"} | status_code >= 500

Metric Queries (PromQL-style)

Metric queries allow you to aggregate and compute over your observability data. These follow PromQL-style syntax.

Selectors

Metric selectors use the same matching operators as log stream selectors:

http_requests_total{method="GET", status="200"}

cpu_usage{instance=~"web-.*"}

Common Aggregation Operators

Examples

sum(rate(http_requests_total{environment="prod"}[5m]))

avg by (service) (cpu_usage)

Range Vectors

Range vectors select a range of samples back from the current instant. Durations are specified in square brackets:

rate(http_requests_total[5m])

Binary Operators

Supported Subset

Lakerunner implements a subset of the full LogQL and PromQL specifications. While we continue to expand coverage, some advanced operators and functions may not yet be available. If you encounter an unsupported expression, reach out to our team and we'll be happy to help find an alternative or prioritize support.

For the complete specification of each language, refer to the official documentation:

• LogQL Documentation (opens in a new tab) — Grafana Loki's query language for logs
• PromQL Documentation (opens in a new tab) — Prometheus's query language for metrics