# See which servers are running hot

> The Capacity & growth report shows CPU, memory and disk for every server over a month you choose — and which disks are filling up.

Source: https://infranest.io/docs/capacity-report
Last updated: 2026-09-30

---

This article is for anyone who needs to spot overloaded servers across their whole fleet before they become a problem.

## Overview

- The Capacity & growth report shows CPU, memory and disk usage for every server over a month you choose, plus which disks are filling up.
- A single server's own page shows this same data just for that server — use the Capacity & growth report instead when you're deciding what needs resizing across your whole Infrastructure.

## Read a row

1. Look at the **peak** figure for the month — this is shown first, with the average underneath it.
2. Check the small line next to the numbers, which shows the shape of the month at a glance.
3. Compare peak and average together: a server that sits at a steady 60% and a server that had one bad afternoon can show the same average, but only the peak tells them apart.
4. Watch the colour of the numbers — they turn amber above 75% and red above 90%.

<!-- screenshot: capacity-report -->

<!-- docs-screenshots:start:capacity-report -->

![See which servers are running hot](/media/70211ff7-bd4f-4afd-a34b-95370cf14b08)

<!-- docs-screenshots:end:capacity-report -->

## Understand "12 of 31 days"

1. Check the days-measured text on each row, such as "12 of 31 days" — this tells you how much of the month the figures are based on.
2. Treat a server measured on fewer days with caution — it isn't directly comparable to one measured every day of the month.
3. Look for a note about when the server was added if it only has a short measurement window — this means it's new, not that monitoring is broken.
4. Watch for rows that say "No readings" — this means the server didn't report in at all, not that it used zero resources.

## Check disk forecasts

1. Look for a line like "full in ~9d" on a disk that's filling on a steady trend.
2. Remember forecasts only appear on a period that includes **Today** — they're based on the last week of readings, so they describe right now, not a past month.
3. Don't expect a forecast on every disk — one-off dumps or a build cache that fills and empties aren't on a steady trend, so no forecast is shown rather than a misleading one.

> [!NOTE]
> A missing forecast isn't an error — it means the disk's usage pattern isn't steady enough to predict reliably.

## Export the report

1. Select **Download** to export the month as shown, laid out as a CSV file.
2. Check the exported file for the days-measured columns, included so the figures can be double-checked.

> [!TIP]
> Every export is recorded in your Audit Log, so you can always confirm who downloaded a report and when.

## Troubleshooting

- **A server shows "No readings" instead of numbers** — this means InfraNest didn't receive any data from that server for the month, not that it was idle. Check the server's monitoring setup.
- **No forecast shown for a disk that's clearly filling up** — this is expected if the usage isn't on a steady trend, or if you're viewing a past month rather than one that includes Today.