loki

To parse unstructured log lines in Loki with LogQL, apply pipeline stages like json, logfmt, pattern, or regexp parsers to extract labeled fields from raw log streams. These parsed fields become available for subsequent filtering and metric aggregation queries.

Filtering Kubernetes logs in Loki requires using label matchers in your LogQL stream selector to target specific service and environment labels, followed by efficient line filters to narrow down the exact log events within those streams.

Calculate request latency percentiles in Loki by parsing the duration field from your logs with a LogQL pipeline stage, unwrapping that numeric value, and applying the quantile_over_time function across a specified range vector interval.

When Loki queries return no log streams, verify that your LogQL label selectors match the labels actively pushed by the distributor and check if the queried time range falls within the ingester retention window. Query frontend behavior also impacts stream visibility.

Yes, Loki integrates with Grafana Alloy to ingest Kubernetes logs via the push API. Alloy collects log streams from your cluster and pushes them to the Loki distributor, making them immediately available for LogQL querying.