Windmill v1.661.0: Native OTel Metrics for Production Observability

Here at industry sources, we tracked Windmill's observability roadmap closely, and v1.661.0 represents a significant shift toward production-grade monitoring. Windmill now ships with native OpenTelemetry metrics support, eliminating the need for custom exporters or middleware to surface workflow execution data. This means builders can push metrics directly to any OTel-compatible backend - Datadog, New Relic, Grafana Cloud, or self-hosted Prometheus - without integration friction.

The implementation surfaces critical workflow signals: execution duration, success/failure rates, queue depth, and resource utilization. For teams running Windmill in production, this shifts observability from post-hoc logging to real-time metric streams. You can now set up alerting on workflow SLOs immediately after deployment rather than waiting for custom dashboards to mature.

• Native OTel metrics export - no agent configuration required
• Captures execution duration, throughput, error rates, and queue metrics
• Works with any OTel-compatible backend without custom code
• Reduces observability setup time from days to minutes
• Enables SLO-driven alerting on workflow performance

Workflow platforms live or die on visibility into execution health. Before this update, Windmill operators typically built custom logging layers or relied on periodic health checks - both lag behind real-time needs. OTel metrics close that gap by exposing workflow behavior in the same observability layer your infrastructure already uses.

More concretely: if you're running Windmill workflows that process payments, handle data pipelines, or trigger downstream systems, you now have a standard way to track whether those workflows are performing within SLOs. You can alert on queue depth before workflows back up. You can correlate workflow metrics with database performance or API latency without manual correlation. This is table stakes for production systems, and Windmill is catching up to what teams expect.

The OTel standard also future-proofs your observability investment. You're not locked into Windmill's metrics format or backend choice. Switching monitoring vendors no longer requires re-instrumentation.

• Visibility into real-time workflow health without custom code
• SLO alerting becomes immediately actionable
• Correlate workflow metrics with infrastructure metrics natively
• Avoid vendor lock-in through OTel standardization
• Reduces toil spent building bespoke dashboards

If you're already running Windmill, activation is straightforward. Configure your OTel exporter endpoint in Windmill's config - most teams point to their existing Prometheus scrape targets or OTel collector. Windmill will begin emitting metrics immediately. If you don't have an observability backend yet, this is a signal to provision one: Grafana Cloud's free tier works for initial exploration.

Start with three dashboards: workflow execution latency by step, failure rate trending, and queue depth over time. These three metrics will surface 80% of production issues. Once those are stable, layer in resource utilization (CPU, memory per workflow execution) and cost tracking if you're billing customers per workflow run.

For teams integrating Windmill into larger platform stacks, OTel metrics enable correlation with other instrumented components. If your API gateway, database, and message queue already export OTel metrics, Windmill workflows now fit seamlessly into that chain. The momentum in this space continues to accelerate.

• Configure OTel exporter endpoint in Windmill config
• Start with execution latency, failure rate, and queue depth dashboards
• Layer in resource utilization and cost tracking after baseline is stable
• Correlate with infrastructure metrics for end-to-end visibility