Mastering Log Noise Reduction: Custom Drop Rules in Grafana Cloud

Introduction: The Cost of Noisy Logs

Every observability team deals with logs that add little value—health check noise, forgotten DEBUG statements, or verbose INFO messages from rarely used services. These logs inflate storage costs and obscure actionable insights, creating a frustrating dilemma: eliminate them without involving cumbersome infrastructure changes. Previously, dropping such logs in Grafana Cloud required manual intervention or configuration changes. Now, with the new drop rules feature in Adaptive Logs (public preview), you can define custom rules to discard low-value logs before they are ingested, saving money and reducing clutter instantly.

This capability extends the same drop logic already available in Adaptive Metrics and Adaptive Traces, but now tailored for log streams. It complements intelligent optimization recommendations by letting you add your own override criteria.

How Drop Rules Work

With each drop rule, you specify logic using a combination of log labels, detected log levels, and line content. The rule is evaluated before logs are written to Grafana Cloud Logs, so dropped lines never incur storage or indexing costs. The system supports a drop percentage (0–100%) for cases where you want to reduce volume but keep a representative sample.

Rules are evaluated in a defined priority order (you can reorder them). The first matching rule applies its drop rate, and subsequent rules are ignored for that log line. This allows you to create layered strategies: for example, drop 100% of DEBUG logs for one service, but only 50% for another.

Check out the practical use cases below for real-world examples.

Practical Use Cases for Drop Rules

1. Drop Logs by Level

Eliminate DEBUG or TRACE level logs that are consuming your logging budget without adding value. Create a rule that drops all DEBUG logs (drop rate 100%). This is ideal for development environments or for services where debug information is irrelevant after a certain period.

2. Sample Chatty, Repetitive Logs

Some logs are too valuable to drop entirely but appear thousands of times per second. Use a drop percentage to sample them. For instance, set a drop rate of 90% to keep 1 out of every 10 occurrences. This retains enough context for troubleshooting while slashing volume by 90%.

3. Target a Specific Noisy Producer

A microservice starts emitting excessive INFO or WARN logs due to a bug or misconfiguration. Instead of asking the team to change their logging framework, you can create a rule that matches by label selector (e.g., service=payment-gateway) combined with log level or a text string (e.g., "healthcheck"). Apply a 100% drop rate to stop those lines entirely.

These examples demonstrate flexibility: rules can be as broad or as granular as needed.

Integration with Exemptions and Patterns

Drop rules are one of three mechanisms in Adaptive Logs that manage log volume. The evaluation happens in a fixed order:

1. Exemptions – Protect critical log lines from any sampling. If a log matches an exemption, it passes through untouched.
2. Drop rules (your custom rules) – Evaluated in priority order. The first matching rule applies its drop percentage.
3. Patterns – Optimization recommendations (based on repetitive patterns) are applied to remaining log lines that were not exempted or dropped.

This layered system gives you complete control. For example, you might exempt security audit logs, drop all DEBUG logs from development services, and then let Adaptive Logs’ pattern recommendations handle the remaining repetitive lines from production. The result: only high-value logs reach your storage.

Also, drop rules work alongside Adaptive Metrics and Adaptive Traces drop rules, providing a consistent governance framework for all telemetry types.

Getting Started with Drop Rules

To begin using drop rules:

• Navigate to Adaptive Logs in your Grafana Cloud instance.
• Create a new drop rule from the “Drop Rules” tab.
• Define your criteria (labels, level, content) and set the drop percentage.
• Order rules by priority (drag and drop).

The feature is in public preview, so no special enablement is required. You can start saving costs immediately.

Internal anchor: Return to use cases

Conclusion

Drop rules in Adaptive Logs provide a simple, powerful way to eliminate wasteful log lines without altering application code or relying on other teams. Whether you need to remove all DEBUG logs, sample repetitive output, or target a specific service, these rules give you immediate control over log volume and cost. Combined with exemptions and pattern-based recommendations, they form a complete log cost management solution. Start exploring the Grafana Cloud documentation to implement your first drop rule today.