How to Create Log-Based Metric in GCP: Step-by-Step Guide

To create a log-based metric in GCP, go to the Cloud Logging console, select Logs-based Metrics, then click Create Metric. Define the metric name, description, and filter to specify which logs to count or measure, then save it to start collecting data.

Syntax

The main parts to create a log-based metric in GCP are:

• Metric Name: A unique name for your metric.
• Description: A short explanation of what the metric tracks.
• Filter: A query that selects which log entries count towards the metric.
• Metric Type: Choose between Counter (counts matching logs) or Distribution (measures numeric values in logs).

terraform

resource "google_logging_metric" "example_metric" {
  name        = "example_metric"
  description = "Counts error logs from my app"
  filter      = "resource.type=\"gce_instance\" AND severity>=ERROR"
  metric_descriptor {
    metric_kind = "DELTA"
    value_type  = "INT64"
  }
}

Example

This example creates a log-based metric that counts error logs from Compute Engine instances.

terraform

resource "google_logging_metric" "error_count" {
  name        = "error_count"
  description = "Counts error logs from Compute Engine instances"
  filter      = "resource.type=\"gce_instance\" AND severity>=ERROR"
  metric_descriptor {
    metric_kind = "DELTA"
    value_type  = "INT64"
  }
}

Output

Apply complete! Resources: 1 added, 0 changed, 0 destroyed.

Common Pitfalls

• Using an incorrect or too broad filter can cause metrics to count unwanted logs or no logs at all.
• Not setting the correct metric_kind and value_type leads to errors or wrong metric behavior.
• For distribution metrics, forgetting to extract numeric values from logs will cause the metric to fail.

terraform

resource "google_logging_metric" "wrong_metric" {
  name        = "wrong_metric"
  description = "Incorrect filter example"
  filter      = "severity=ERROR"  # Missing resource type causes too broad matching
  metric_descriptor {
    metric_kind = "DELTA"
    value_type  = "INT64"
  }
}

# Corrected filter:
resource "google_logging_metric" "correct_metric" {
  name        = "correct_metric"
  description = "Counts error logs from GCE instances"
  filter      = "resource.type=\"gce_instance\" AND severity>=ERROR"
  metric_descriptor {
    metric_kind = "DELTA"
    value_type  = "INT64"
  }
}

Quick Reference

Remember these key points when creating log-based metrics:

• Use precise filters to target relevant logs.
• Choose DELTA for counting logs and DISTRIBUTION for measuring values.
• Give clear names and descriptions for easy identification.
• Test your filter in the Logs Explorer before creating the metric.

Key Takeaways

Create log-based metrics in GCP by defining a name, description, and filter in Cloud Logging.

Use precise filters to count or measure only the logs you need.

Choose the correct metric type: DELTA for counts, DISTRIBUTION for numeric values.

Test filters in Logs Explorer before creating metrics to ensure accuracy.

Avoid broad filters and incorrect metric descriptors to prevent errors.