Promql Custom Metric Otlp Monitoring Api Alert

What Cloud Monitoring PromQL Metric Mapping Is

When querying metrics ingested through the Cloud Monitoring API (googlecloudmonitoring ingestion mode) using Prometheus Query Language (PromQL), metric names must follow specific conversion rules to be compatible with PromQL syntax. In Cloud Monitoring, the metric type is converted into a PromQL identifier, and monitored resources must be explicitly defined where applicable.

How It Meets the Requirements

1. Metric Name Transformation: The standard Cloud Monitoring metric mapping rules mandate:
   • The first forward slash (/) in the metric type string is replaced with a colon (:).
   • All other special characters, including periods (.) and subsequent forward slashes (/), are replaced with underscores (_).
   • Therefore, workload.googleapis.com/custom/http_errors_total becomes workload_googleapis_com:custom_http_errors_total.
2. Monitored Resource Filtering: Because metrics ingested via the Cloud Monitoring API can theoretically attach to multiple resource types rather than defaulting to a single Prometheus target, PromQL queries must explicitly include the {monitored_resource="gce_instance"} label filter.
3. Rate and Ratio Calculation: Wrapping the cumulative counters in rate(...[5m]) calculates the per-second rate of change over a 5-minute window, and computing the ratio with / compared against > 0.05 enforces the 5% threshold.

Key Advantages and Capabilities

• Syntactic Compliance: Ensures query validation succeeds in the Cloud Monitoring PromQL editor.
• Fleet Aggregation: Using sum() aggregates request and error rates across all matching VM instances.
• Precise Scoping: Explicitly filtering on monitored_resource="gce_instance" avoids ambiguity in multi-resource environments.

Why This Is Optimal

This syntax precisely follows Google Cloud's documented mapping specification for non-Prometheus native custom metrics ingested via the Monitoring API, avoiding runtime query errors and missing data anomalies.