---
title: "Check which time window an alert investigation used | Proofline Help"
description: "Open Evidence and compare the symptom window with the metric condition and the alert’s opening time. Those times can differ for long-running or missing-data conditions."
url: https://proofline.sh/help/investigation-symptom-window/
---

[Help center](/help/)

# Check which time window an alert investigation used

Open Evidence and compare the symptom window with the metric condition and the alert’s opening time. Those times can differ for long-running or missing-data conditions.

## Compare the metric window with symptom onset

When an alert's condition covers a longer period than the alert's own symptom, such as a daily total, or opens only after its metric reported nothing for more than 15 minutes, the symptom starts that period before the alert opened, at most 24 hours before. The timeline then marks, for example, **24-hour symptom period began**, the **Symptom window** covers that period, and Slack times leads against it, such as "deployed 19h after the 24-hour symptom period began". For an alert that opened after such a silence, when Proofline's metric read finds a value inside the silence, the leads are ranked against the alert's own symptom window instead.

In **Evidence**, check the **Symptom window**. The window starts where Proofline assumed the symptom began. When the alert's own metric read shows that the metric already met the alert's condition before then, the row says from when, such as **The alert's metric already met its condition from Nov 2, 8:15 AM, 30m before this window.** The read can also show the reverse: the metric first met the condition after the window began, and the row then says when, such as **The alert's metric first met its condition at Nov 2, 9:00 AM, 15m after this window began.** For an alert that counts events over its period, such as log entries, the row gives the first minute that counted anything, such as **The alert's metric first counted at Nov 2, 9:30 AM, 45m after this window began.** That minute can come before the count reached the threshold. Either way the window still starts where Proofline assumed the symptom began; the time says how much of the window passed before the metric showed anything. Without that read, or when it shows none of these, the row adds nothing. For an alert investigation, the next row names the alert's source, such as **CloudWatch alarm**, and opens the alert with **Open in Cloud Monitoring**, **Open in CloudWatch**, **Open in Sentry** or **Open in Azure Monitor** when Proofline can link to it. Without a link, a Cloud Monitoring or Sentry alert's row shows the alert's recorded identifier, and a CloudWatch alarm or an Azure Monitor alert has no such row. **Reviews Proofline read** links to the reviews used by the investigation. A review that counts only for its environment findings from Kian is not listed there; its timeline entry and the lead it ranks cite it. Open a review's **Receipt** to see exactly which review Proofline read: when it was recorded, in UTC, and its **Commit SHA**, **Evidence SHA-256**, **Bundle ID**, **Assessment ID** and **Repository ID**, each with a copy button. A newer review of the same commit does not replace it. For a question, **Related alerts** links each alert investigation the question used, with its trigger and when Proofline received it. When an alert gave the question no telemetry, its row says **Had no result when the question read it** or **Its result held no telemetry**. [Read the answer](/help/ask-for-investigation/#read-the-answer) explains which alerts a question uses.

## Related guides

-   [Find an environment's investigations](/help/environment-investigations/)
-   [Create an environment](/help/create-environment/)
