Log-based metrics in GCP - Commands & Configuration

Introduction

Sometimes you want to count or measure things happening in your logs automatically. Log-based metrics let you turn log messages into numbers you can watch and alert on.

When you want to count how many errors happen in your app from logs.

When you want to measure how often a specific event appears in logs over time.

When you want to create alerts based on unusual log activity.

When you want to track performance or usage by analyzing log entries.

When you want to visualize log data trends in dashboards.

Config File - log_metric.yaml

log_metric.yaml

apiVersion: logging.googleapis.com/v2
kind: LogMetric
metadata:
  name: error_count_metric
spec:
  description: "Count of error logs"
  filter: "severity=ERROR"
  metricDescriptor:
    metricKind: DELTA
    valueType: INT64
    unit: "1"

This file defines a log-based metric named error_count_metric that counts how many log entries have severity ERROR.

filter selects which logs to count.

metricDescriptor describes the metric type and value.

Commands

This command creates a log-based metric named error_count_metric that counts log entries with severity ERROR.

Terminal

gcloud logging metrics create error_count_metric --description="Count of error logs" --filter="severity=ERROR" --metric-descriptor-metric-kind=DELTA --metric-descriptor-value-type=INT64

Expected Output

Created [https://logging.googleapis.com/v2/projects/my-project/metrics/error_count_metric].

→

--description - Describes what the metric counts

→

--filter - Selects which log entries to include

This command lists all log-based metrics in the project to verify the new metric was created.

Terminal

gcloud logging metrics list

Expected Output

NAME                DESCRIPTION

error_count_metric  Count of error logs

This command deletes the log-based metric when it is no longer needed to keep metrics clean.

Terminal

gcloud logging metrics delete error_count_metric

Expected Output

Deleted [https://logging.googleapis.com/v2/projects/my-project/metrics/error_count_metric].

Key Concept

If you remember nothing else from this pattern, remember: log-based metrics turn log messages into numbers you can watch and alert on automatically.

Common Mistakes

Using a filter that matches no logs or too many logs

The metric will not count anything or will count irrelevant logs, making it useless or noisy.

Write a precise filter that matches only the log entries you want to measure.

Not verifying the metric after creation

You might think the metric works but it may not be created or configured correctly.

Always list and check your metrics after creating them to confirm.

Summary

Create a log-based metric with gcloud logging metrics create and a filter to select logs.

List metrics with gcloud logging metrics list to verify creation.

Delete metrics with gcloud logging metrics delete when no longer needed.