Loghouse as a source for Prometheus

[z9r5]

Loghouse queries can be used as metrics for Prometheus. Consider query select count(*) as value, namespace from logs where log ~ ".*error.*" group by namespace. This query can be run every minute, so Prometheus will get errors number as a metric: log_errors{namespace="$namespace"} $value.

Features:

• metrics should have types: gauge, counter, histogram;
• metrics’ labels should be populated from query’s results or parameters;
• metrics’ value should be populated from query’s results;
• loghouse-backend should have metrics endpoint for Prometheus scrapers.