Capture Diagnostics on an Interval
Automatic recapture takes the same diagnostics report over and over on a fixed interval, replacing the one on screen each time. Instead of a still photograph of the engine, you get a live feed of it.
That difference matters for a specific class of question. A single capture can tell you that a load has read 4 million rows. Only a sequence of captures can tell you whether that number is still climbing. Use automatic recapture whenever the answer depends on change over time rather than on a single moment.
Permissions
Only users with the Admin role can open the Diagnostics subtab or capture a report. See the Permissions page for the full matrix.
When to Use It
| SITUATION | WHAT WATCHING OVER TIME TELLS YOU | WHAT TO WATCH IN THE REPORT |
|---|---|---|
| An operation looks stuck and you need to know whether it is slow or hung. | Whether the counters move at all between captures. Frozen counters mean hung; slowly rising counters mean slow. | Running loads, specifically Rows read and the Input advanced column. |
| You just restarted the Dataplane. | Whether queued work is actually being picked up and drained. | The Workload tiles for task demand, and Active work for container progress. |
| You raised capacity or a scaling event is under way. | Whether new executors are arriving and being used, or the cluster is churning. | Executor stability for added and removed events, and the executor capacity tiles. |
| A finding appeared and you are unsure whether it is real. | Whether it persists across captures or was a momentary blip. | The Findings table, watching whether the row survives and how its Since age grows. |
| An incident is developing and you want to see it unfold. | The sequence of events rather than the aftermath. | The whole report, refreshed every few seconds. |
| A source is intermittently slow. | Whether latency spikes come and go, which a single capture would miss entirely. | Source latency, watching p95 move between captures. |
Tip
Pick the interval to match what you are watching. 5s is for a live incident you are staring at. 30s is for confirming that a long recovery is trending in the right direction without hammering the engine.
How It Differs From a Manual Capture
Automatic recapture produces exactly the same report as a manual capture. Only the trigger and the on-screen feedback differ.
| Manual recapture | Automatic recapture | |
|---|---|---|
| Trigger | You click the button. | A timer, at the interval you chose. |
| Feedback while capturing | The button shows a loading state. | None. The new report swaps in quietly. |
| If the capture fails | An error notification appears. | It fails silently and the report on screen stays. |
| Report contents | Identical. | Identical. |
For what those contents are, section by section, see Dataplane Diagnostics. To keep any single one of them, see Download a Diagnostics Report.
Steps
Step 1: With the Diagnostics subtab open, click the Auto recapture button next to the recapture button.
Step 2: Select an interval: 5s, 10s, or 30s. The chosen label appears next to the icon and the card header switches to Auto recapture active.
Step 3: A report is captured right away, then again on every interval. Each one loads quietly in the background, and the report on screen stays visible and holds its scroll position until the replacement is ready.
Step 4: To stop, open the same menu and select Off.
Note
You can still click Recapture diagnostics while an interval is running. Doing so captures immediately and restarts the countdown to the next automatic capture.
Note
Switching to the Configuration subtab pauses automatic recapture, and switching back resumes it at the same interval. Leaving the Status page turns it off entirely: the panel always starts at Off the next time you open it.
Warning
Every capture makes the engine render a report while it is running your workload. Use the short intervals to watch a live incident, then turn automatic recapture back off.