Windmill Adds OpenTelemetry Metrics: What Builders Need to Know

Here at industry sources, we track infrastructure updates that shift how builders approach observability. Windmill v1.661.0 introduces native OpenTelemetry metrics support, meaning your Windmill deployments can now emit standardized metrics compatible with any OTel-compliant backend - Prometheus, Datadog, New Relic, or your own stack.

This isn't a surface-level integration. Windmill now exposes workflow execution metrics, task duration distributions, error rates, and resource consumption as proper OTel instrumentation. You get histogram and counter metrics out of the box, not post-hoc logging analysis.

The implementation allows you to configure which metrics to export and where to send them. This gives you control - you're not forced into a particular monitoring vendor or overwhelmed with noise.

• Native OTel metrics export with zero custom code required
• Histogram and counter metrics for workflow execution and task performance
• Compatible with any OTel-compliant observability backend
• Configurable metric export to avoid vendor lock-in
• Reduces dependency on application-level logging for performance analysis

If you're running Windmill in production, observability isn't optional - it's how you detect degradation before it becomes an outage. Before v1.661.0, you were limited to logs and custom dashboards. Now you have structured metrics at the platform level.

This is critical for SLO tracking. Workflow execution latency, task failure rates, and queueing metrics are now queryable and alert-able. You can set meaningful thresholds and catch performance regressions immediately, not hours later.

For multi-tenant or high-volume deployments, OTel metrics also reduce cardinality problems. Instead of parsing log timestamps and calculating durations yourself, you get pre-aggregated histograms that play nicely with cost-conscious observability stacks.

• Enables SLO-driven monitoring without custom instrumentation
• Reduces time-to-insight for performance issues
• Lowers cardinality cost on observability platforms
• Integrates with existing observability infrastructure
• Provides audit trail for workflow execution patterns

Deployment is straightforward. After upgrading to v1.661.0, you configure the OTel exporter endpoint in your Windmill deployment. For Kubernetes users, this is an environment variable or ConfigMap entry. For self-hosted setups, it's a configuration parameter in the Windmill config file.

The integration model matters here. If you're already using OTel collectors for other services, Windmill metrics flow into the same pipeline. If you're not, this is a good forcing function to establish observability discipline. Start with Prometheus and Grafana, or use a managed backend like Datadog.

One architectural note: OTel metrics are separate from logs. You still get structured logs for debugging, but now you also have metrics for monitoring. This separation of concerns makes alerting cleaner and reduces false positives from noisy logs.

The momentum in this space continues to accelerate.

• Configuration via environment variables or config files
• Works with existing OTel collector infrastructure
• Separates monitoring metrics from debug logs
• No performance overhead for opt-in metric export
• Pairs naturally with Prometheus, Datadog, and similar platforms