Monitoring

By the end of this page you will know where to watch a deployment’s traffic, trace individual requests, review agent runs, and read conversations.

Prerequisites

• role or higher.
• A running deployment with some traffic.

Observability

Open a deployment and select Observability. Pick a time range (last hour, 6 hours, 24 hours, or 7 days) and switch between the Summary, Traces, and Activity views. Changing the range refetches the data.

Summary

The Summary view charts the deployment’s traffic and health for the selected range: request count, error rate, latency percentiles, an apdex score, and — for deployments that run on a dedicated pod (Web App and Copilot) — pod and resource health.

Traces

The Traces view lists individual requests, filterable by path, method, and status. Open one to see its full waterfall — each span, how long it took, and where time went.

Activity

The Activity view is the deployment’s audit log — deployment, pod, gateway, and configuration events, each with a severity. It’s the first place to look when provisioning or an update didn’t go as expected.

Executions

Select Executions to see the deployment’s agent runs — status, mode (for example, a scheduled or API run), token usage, tool and LLM calls, duration, and cost. Open a run to see its detail, including its span waterfall.

Conversations

For deployments that hold conversational agents, select Conversations to review chat sessions: totals for sessions, messages, and unique users, a filterable list, and a full transcript of any session (user, assistant, and tool messages).

Logs

Deployments that run on a dedicated pod — Web App and Copilot — have a Logs view that streams the running app’s logs live.

Common issues

• No data yet — metrics and traces appear once the deployment receives traffic. Send a request or run a test.
• Resource metrics say “not available” — resource charts need cluster metrics; they may be absent in some environments even though the deployment is healthy.