Skip to content

Release confidence

Review the impact of a release

See whether a promoted version improved latency, increased errors, or changed resource use. Keep the evidence alongside the release outcome.

Observe both sides of the release

The monitor collects a baseline and an observation window. The report compares latency, errors, throughput, CPU, memory, restarts, and health failures with the samples behind each value. Sample counts show whether the comparison has enough evidence to interpret, and unavailable measurements remain distinct from a normal result.

Observe both sides of the release: The monitor collects a baseline and an observation window

This workflow is useful when a deployment completes but you still need to know how it affected the service. Allow representative traffic before and after the release, then review the time windows and sample counts. The comparison can expose changes in customer experience that a startup or readiness check would not necessarily show.

Observe both sides of the release: This workflow is useful when a deployment completes but you still need to know how it affected the service

Recognize regressions and missing evidence

The metric verdict is separate from promotion or rollback. Insufficient samples and unavailable metrics stay explicit; an unmeasured release is never given a perfect impact score. Review the per-metric evidence to understand what actually contributed to the reported verdict before making an operating decision.

Recognize regressions and missing evidence: The metric verdict is separate from promotion or rollback

An increase in errors or latency deserves investigation even when the deployment itself succeeded. At the same time, changing demand can explain a throughput difference without indicating a fault. Review the individual measurements and their evidence before relying on the combined verdict, and keep unavailable values visible instead of interpreting them as healthy results.

Recognize regressions and missing evidence: An increase in errors or latency deserves investigation even when the deployment itself succeeded

Share the same report across the team

Read the report offline on the host or review its synchronized dashboard counterpart. Analysis informs your operating decision and never changes deployment verification or rollback behavior. Keep the report with the release review so the person taking over has the same evidence and understands the decision.

Share the same report across the team: Read the report offline on the host or review its synchronized dashboard counterpart

Use the shared report during handover so the next person sees the same release, observation window, and metric evidence. If you decide to restore a version, check whether the affected customer operation recovers afterward. The report helps explain the decision but does not automatically execute a rollback or reverse changes in external systems.

Share the same report across the team: Use the shared report during handover so the next person sees the same release, observation window, and metric evidence

One binary between git push and production.

Install the CLI, verify the host, and follow the quick start to build your first application.

Read the docs