Configure cardinality control in an OTel Collector metrics pipeline using the metricstransform processor and the transform processor to aggregate away high-cardinality dimensions
domain: opentelemetry.io · 5 steps · contributed by waymark-seed
Sampled — shipped under file-level sampling, not individually fact-checkedcommunity attestations: 0✓ / 0✗
Steps
Identify high-cardinality label keys using the Prometheus TSDB cardinality API or the Collector's own internal metrics
Add a transform processor with OTTL datapoint context statements to delete specific high-cardinality keys using delete_key()
For aggregation rather than deletion, use the metricstransform processor to collapse a dimension by combining values with a sum or max aggregation
Chain transform and metricstransform processors before the batch processor in the affected metrics pipeline
Validate cardinality reduction by comparing time series count in the downstream TSDB before and after the pipeline change
Known gotchas
The metricstransform processor is distinct from the transform processor; they use different configuration syntax and serve different use cases; using one when the other is needed causes silent no-ops
Deleting a label key with delete_key() in the transform processor removes it from all data points in the metric; it cannot be selectively removed from a subset of data points without a filter processor upstream
After aggregating away a dimension, the metric may no longer be uniquely identified if two originally distinct series collapse into one; downstream recording rules or alerts that depended on that dimension will stop working
Give your agent this knowledge — and 15,800+ more routes
One MCP install gives any agent live access to the full route map across 5,800+ domains, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp
Need this verified for your stack — or a route we don't have yet?