Skip to main content
In Mage OSS, use the Overview page to monitor aggregate pipeline run health across your project. In Mage Pro, select Monitoring in the left navigation. The monitoring panels show recent run status, failing runs, active runs, and trigger activity for the selected time range. In Mage Pro, the global Monitoring page supports rolling monitoring ranges for Past 24 hours, Past 7 days, and Past 30 days, plus a Custom UTC range. The selected range drives the overview metric tiles, run-health chart, failed runs, active runs, longest active runs, and AI Sidekick prompts. The global Monitoring page does not include an average-duration tile or completed-run duration trends; those runtime metrics remain available on individual pipeline monitoring pages. Custom ranges require the end time to be after the start time before they can be applied, and the selected range is preserved through refreshes and relevant navigation. For a custom range in Mage Pro, edit the UTC date and time segments directly in each YYYY-MM-DD HH:mm input, or choose them in the calendar and time picker. The input keeps separators in place and shows placeholders for missing digits. Complete valid values and picker selections stay in sync; incomplete entries or an invalid range prevent the range from being applied. Monitoring data availability can further limit the dates you can select. In Mage Pro, the Longest active pipeline runs chart at the bottom of the Monitoring page ranks non-streaming pipeline runs by elapsed runtime, with the longest run first. It shows five runs by default; use the Show selector to display 5, 10, 25, or 50 runs. The ranked list scrolls within the panel when needed. Each row shows the pipeline name, run ID, trigger name when available, and elapsed duration. Streaming pipelines are excluded. Durations refresh automatically, and selecting a pipeline name opens that run’s details. If the initial request fails, use Retry in the panel. If a later refresh fails, the panel keeps the last loaded runs visible and shows a warning with Retry.

​
Metrics

Here are the following metrics that are monitored for each pipeline and block:
  • Pipeline runs completed successfully
  • Pipeline runs failed
  • Block runs completed successfully
  • Block runs failed
  • How long each block ran for
  • CPU usage for Kubernetes block and pipeline runs in Mage Pro
  • Memory usage for Kubernetes block and pipeline runs in Mage Pro
  • Data validation success for each block (coming soon)
  • Data validation failure for each block (coming soon)
In Mage Pro, the pipeline monitoring page includes runtime trends for performance debugging. Pipeline runtime over time shows the average duration of successfully completed pipeline runs, while Block runtime by block lets you select a block and chart the average duration of its completed samples. The existing Runtime hot spots list remains useful for finding blocks with the highest average and maximum runtimes before selecting one for its time-series breakdown. Runtime trends use the dashboard’s selected UTC time range and trigger filter. The Past 24 hours view groups samples by hour; longer ranges group them by date. A chart needs completed samples in at least two time buckets, so a block or pipeline with fewer populated buckets shows an empty state instead of a trend line. If runtime data fails to load, use Retry in the affected chart. For pipelines that run with the Kubernetes executor in Mage Pro, the pipeline monitoring dashboard can show historical peak CPU and memory usage, sample counts, and copied resource request and limit recommendations based on stored run metrics. See Monitoring with Prometheus to set up Prometheus as the preferred resource usage source.

​
Monitoring actions

The Monitoring page in Mage Pro includes actions for the monitoring panels. Fix runs and Analyze runs are available when AI Sidekick is enabled.
  • Fix runs opens AI Sidekick with a prompt to investigate failing runs for the selected time range.
  • Analyze runs opens AI Sidekick with a prompt to inspect active runs for the selected time range.
  • View triggers opens the trigger list, preserving the selected compute cluster filter when one is active.

​
Streaming health

Streaming pipelines in Mage Pro include a Streaming health view on the pipeline monitoring page. Use this view to check whether a running or recent stream is still accepting source records and whether Mage is writing those records to each destination. When no time range is specified in the URL, Streaming health opens to Last day. The Streaming health view shows:
  • Rows consumed from the streaming source.
  • Destination writes accepted by Mage, broken down by destination.
  • Failed writes for destinations that reported write errors.
  • Average processing latency when latency samples are available.
  • The last metric update time and the time bucket size.
The throughput chart compares source consumption with destination writes over time. Source consumption is shown separately from destination writes because destination totals can be higher than consumed rows when a stream writes to more than one destination. For longer selected time ranges, Mage Pro rolls chart points into coarser time buckets before returning Streaming health metadata so the chart stays responsive while preserving the selected range totals. Destination rows include each destination block, its exported count, failed write count, and last successful write time. On the pipeline monitoring page, streaming runs are grouped by trigger. This is useful for long-running streaming triggers because Mage can create a replacement pipeline run for the same trigger after an exception. Trigger-level health keeps those automatic restarts visible as one operational stream.