Read an incident
The InsightX page has the incident list on one side and the selected incident on the other. The newest incident is selected when you open the page.
The incident list
Each row in Incidents shows the incident title, when it was detected, its status, and a one-line summary. When the incidents come from more than one resource, the row also shows which resource it belongs to.
Select a row to read the incident. Use Refresh to load new incidents.
Filters
Two filters sit above the list:
- Filter by resource: All resources, or one resource of your service.
- Filter by date: All time, Last 24 hours, Last 7 days, or Last 30 days.
If nothing matches, the page shows No incidents match these filters. Select Clear filters to see everything again.
Opening InsightX from View in InsightX on the Resources tab starts with the resource filter already set.
Statuses
| Status | Meaning |
|---|---|
| Explained | InsightX finished and the incident has a full explanation. |
| Investigating | InsightX is still looking into the incident. The page updates by itself every few seconds until the explanation is ready. |
| Couldn't analyse | InsightX could not finish its analysis. The incident was still detected and handled. If the page shows Try again, you can ask for the analysis to be run again. |
| Detected | The incident was recorded without a detailed analysis. The page shows "Detailed analysis isn't available for this incident." |
The incident
The top of an incident shows its title, the time it was detected, its status, and a few details:
- Exit code: the code your app exited with, when it crashed.
- commit: the commit that was running, when it applies.
- confidence: how sure InsightX is of its explanation. Incidents that Kuberns measured, such as a full disk, have no confidence figure because a measurement is not an estimate.
- ✓ Recovered automatically: your app came back on its own.
- Needs attention: the incident has not recovered and has no detailed analysis.
Below that, an explained incident has these sections:
| Section | What it tells you |
|---|---|
| What happened | A short, plain description of the incident. |
| Why it happened | The reason, with the Root cause on its own line. |
| Impact | What it affected, and how serious it is. |
| What to do next | The steps to fix it. |
| Evidence | The measurements or log lines the explanation is based on, each with its source. |
| What InsightX checked | The lookups InsightX made while investigating, such as reading your recent logs. Shown for crashes and failed health checks. |
Copy an incident for your AI assistant
Select Copy this incident for your LLM to copy a report of the incident. Paste it into an AI assistant to get help fixing the cause. The report has an introduction for the assistant, the incident details, and every section of the explanation.
Try again
If an analysis fails or gets stuck, an incident can show a Try again button. It asks InsightX to analyse the same incident again. This is available for 24 hours after the incident, a limited number of times, and only when the incident has not already been explained.
When there are no incidents
A service with nothing to explain shows Nothing to explain yet. When something goes wrong with your service, the incident appears here.