Metrics Screen

Open Metrics from the main navigation to examine activity in the current Site. Use the charts to identify periods of high concurrency, increased failures, or logging pressure, then use Workflow Logs to investigate individual runs.

Select a time range

Use the time selector at the top of the screen to choose the period to inspect and apply the selection. The default view covers the past 15 minutes. Preset ranges also cover the past hour, 24 hours, seven days, and 30 days.

A longer period uses a coarser time resolution. Compare the same interval when relating a chart to an incident or another chart. The screen displays Site-level samples; it does not identify the failing Step or replace the detailed evidence in a Workflow's logs.

Activity charts

Chart What it shows
Concurrent Workflows (total) Reported Workflow instances running concurrently in the Site. Use it to identify sustained usage and peaks.
Concurrent Workflows (by mode) The reported HTTP, Always On, and other invocation categories. These categories describe the recorded modes, rather than a list of every v2 Runtime trigger type.
Workflow activations Top-level Workflow starts and activations ending with errors.
Sub-Workflow activations Child Workflow starts and activations ending with errors.
Node activations Node invocation starts and invocations ending with errors.

A rise in activations does not necessarily mean a rise in concurrency. Many short runs can finish within the same interval, while a listener or an Always On Workflow may remain active for a long period. Use How Concurrent Workflows Are Measured to understand the capacity rules for the v1 Runtime and the v2 Runtime.

The charts do not provide a separate remaining-capacity counter for the v2 Runtime authoring pool. Do not subtract a by-mode series from a subscription allocation to infer how many Debug slots are available.

Advanced charts

Expand advanced below the activity charts to inspect:

  • Log writes, which separates writes to the Workflow log table, text index, and large-value store.
  • Peak log backlog, which shows reported logging backlog over the selected period.

A backlog can help explain why log information appears later than the activity that produced it. Correlate it with the run time and the logs you are examining; the chart alone does not establish why an individual Property value is missing.

Investigate a change in activity

  1. Select a period that includes the start of the problem and a preceding period for comparison.
  2. Compare total concurrency with activation and error counts.
  3. Open Workflow logs for the same Site and time range.
  4. Identify the affected Workflows and inspect their Step results, errors, and duration.
  5. Review Cluster health or provider behavior when the Workflow evidence points to those dependencies.

For example, if errors increase while concurrency stays low, inspect the failed runs before assuming that capacity is exhausted. If concurrency remains high while few new runs complete, look for long-running Workflows and slow external operations.

See also