Metrics Drop Rules

Drop Rules let you prevent specific metric time-series from being ingested into Oodle. Every sample that matches a drop rule is silently discarded at ingest time, reducing your active series count and storage cost without any change to your exporters or collection pipeline.

Getting Started

Navigate to Metrics → Drop Rules in the sidebar. The list page shows all existing drop rules. Click Create Drop Rule to add a new rule.

Drop Rules Table

Column	Description
Name	Human-readable rule name. Click to open the rule editor.
Metric Name	The __name__ matcher that selects which metrics to drop (e.g. __name__=~"go_gc_.*"). Supports exact match (=), not-equal (!=), regex (=~), and negative regex (!~).
Filters	Optional label matchers that further restrict which series are dropped (e.g. job="unused-exporter").
Blocked Samples/Min	A sparkline trend plus the average number of samples per minute being blocked by this rule over the last hour.
Actions	Context menu with Edit, Clone, and Delete actions.

You can select multiple rules with the checkboxes for bulk deletion. The search bar filters rules by name.

Creating a Drop Rule

1. Click Create Drop Rule on the list page.
2. Rule name — Give the rule a descriptive name (e.g. "Drop unused go_gc metrics").
3. Metric name matcher — Configure the __name__ label matcher:
   • Match type — = (exact), !=, =~ (regex), or !~ (negative regex).
   • Value — The metric name or pattern to match.
4. Additional filters (optional) — Add label matchers to scope the rule further. For example, drop a metric only in a specific cluster or namespace.
5. Click Save to activate the rule.

caution

Drop rules take effect immediately. Any matching samples received after the rule is created will be discarded. The rule does not delete historical data — it only prevents future ingestion.

Editing and Cloning

• Edit — Click the rule name or choose Edit from the context menu to modify the matchers.
• Clone — Choose Clone from the context menu to create a copy of the rule, which you can then modify before saving.

Common Use Cases

Drop all metrics from a specific exporter

Set the metric name to =~ with a prefix pattern like cadvisor_.* and optionally add a filter for job="cadvisor".

Drop a single high-cardinality metric

Set the metric name to = with the exact metric name, such as http_request_duration_seconds_bucket.

Drop metrics only from non-production

Set the metric name as desired, then add a filter like env!="production" to keep production data intact.

Best Practices

• Audit with Metrics Analyzer first — Use the Metrics Analyzer to confirm the metric is unused in dashboards and alerts before dropping it.
• Start with regex — If you are unsure of the exact metric names, use a regex matcher (=~) to catch related metrics.
• Monitor the "Blocked Samples/Min" column — After creating a rule, check back to confirm it is matching the expected volume.
• Name rules descriptively — Include the metric name or purpose so the team can understand each rule at a glance.

Related Pages

• Metrics Analyzer — Identify high-cardinality and unused metrics.
• Metrics Explore — Query metrics with PromQL.