SupportConsole
llms.txtSign in

Read your server's metrics and charts

llms.txt

Every server that SrvYard manages is measured once a minute. The numbers end up on the server's Monitoring tab as a row of tiles for "right now" and a set of charts for "how has it been". You do not install an agent: SrvYard connects over SSH, reads the figures from the operating system and disconnects again.

This article shows what each tile and chart means, how to change the period you are looking at, how old the data can be before SrvYard calls it stale, and how the same figures appear on the server overview. When you know what normal looks like for a server, you can set up an alert rule to tell you when it stops being normal.

Before you begin

  • A server in the Ready state. Samples are only collected while the server is ready. A server that is still provisioning, or one that SrvYard cannot reach, keeps the last sample and shows a notice instead. See Connect an existing server if yours is not ready yet.
  • View access to the server. Anyone who can see the server can see its metrics. Changing monitors needs more: see Alert thresholds and silencing.
  • A few minutes of patience on a new server. A brand new server shows "Collecting the first samples" until the first reading arrives, normally within a minute or two.

1. Open the Monitoring tab

Open the server from Servers in the sidebar and choose Monitoring in the row of tabs under its name.

The Monitoring tab of a server with its tiles
The Monitoring tab: how fresh the data is and the latest number for each resource.
  1. Monitoring. The tab itself. The page title repeats what it does: resource usage sampled over SSH once a minute, with alerts when it crosses a threshold.
  2. Freshness badge. Reads Live with a green dot when the newest sample is less than three minutes old, and Stale with an amber dot when it is older. It also says how long ago the sample was taken, for example "sampled 18m ago".
  3. CPU tile. The share of processor time in use, with a bar and the number of cores. The same tile pattern is used for Memory (used and total, plus swap), Disk (used and total on the main volume) and Load average.
  4. Uptime tile. How long the server has been running since its last boot. When the operating system says a restart is needed after an update, the tile adds A reboot is required, and the server header shows a Reboot required badge.
  5. Updates tile. How many packages can be upgraded. Apply them from the server's Packages tab: see Install system packages and apply updates.

The two remaining tiles, Inbound and Outbound, show the network traffic received and sent, in bytes per second, so you can tell a quiet server from one that is serving a lot of traffic.

Note

A stale badge does not mean the server is down. It means SrvYard has no fresh reading: the server may be busy, restarting, or temporarily unreachable over SSH. Open the server overview to see whether SrvYard still has a connection.

2. Choose the period you want to see

Scroll to Usage over time. The buttons at the right-hand side choose how far back the charts reach.

The range buttons above the CPU and load charts
The range buttons change every chart at once.
  1. Range buttons. 1h, 6h, 24h, 7d and 30d. The page opens on 24h. Every chart on the page changes together, so you can compare CPU, memory and network for the same period.
  2. CPU chart. The percentage of CPU in use over the chosen period, from 0 to 100 per cent. The number at the top right of the card is the latest value.
  3. Load average chart. The one-minute load average. A load around the number of cores means every core is busy; a load far above the number of cores means work is queueing up.

Longer ranges are averaged so that the chart stays readable. For one hour you see one point per minute; the six-hour chart uses five-minute points, the 24-hour chart fifteen-minute points, the seven-day chart hourly points and the 30-day chart one point every six hours. A short spike therefore looks smaller on a long range than on a short one. If something happened at a specific time, look at 1h or 6h around it.

Tip

Start with 7d when you are deciding whether a server is big enough. Daily patterns (a nightly backup, a traffic peak after a newsletter) repeat and show up clearly there. Use 1h when you are watching the effect of something you have just done, such as a deployment or a restart.

3. Read the memory, disk and network charts

Below the CPU and load charts sit three more.

The memory, disk and network charts
Memory (with swap), disk and network traffic over the same period.
  1. Memory. Two lines: Memory is the share of RAM in use and Swap is the share of swap space in use. A server that is working properly keeps swap close to zero. Swap that keeps growing means the server has run out of RAM and is moving data to disk, which makes everything slower.
  2. Disk. The share of the main disk that is full. This line usually climbs slowly. A line that climbs steeply points at logs, uploads, database dumps or old releases piling up.
  3. Network. Inbound and Outbound traffic as two lines on one chart.

Hover over a chart to see the exact value at that moment.

4. See the numbers on the server overview

You do not have to open the Monitoring tab for a quick look. The Overview tab has a Usage card with the same latest numbers.

The Usage card on the server overview
The server overview shows the newest numbers and links to the full charts.
  1. CPU. The number of cores and the percentage in use.
  2. Disk. Used and total space with a bar. Memory sits between the two.
  3. Open monitoring. Takes you to the charts described above. The card also tells you how long ago the figures were sampled.

What SrvYard keeps

  • Sampling. One reading a minute per ready server. At most 20 servers are sampled at the same moment across SrvYard; a server that finds no free slot skips that minute, so you may see a small gap in a busy period.
  • Retention. One-minute samples are kept for 30 days and then deleted. That is why the longest range is 30d: older data does not exist.
  • What is read. Resource figures only: CPU, memory, disk, load, network, uptime and the number of pending updates. File contents and application data are never read.

Troubleshooting

  • The page says "Not collecting right now". The server is not in the Ready state (for example it is provisioning or disconnected). The last sample is shown below the notice. Samples start again when the server is ready.
  • The page says "Collecting the first samples". The server is new and no reading has arrived yet. Wait a minute or two and the tiles appear. If nothing arrives after several minutes, check that the server shows Ready on its overview.
  • The badge says Stale. The newest reading is more than three minutes old. Check that the server is reachable (the overview shows when SrvYard last connected). A server under very heavy load, or one that is rebooting, can miss readings for a while.
  • The memory tile is high but the server is fine. Linux uses free RAM as a file cache, so a long-running server often shows a lot of memory in use. Watch the Swap line instead: swap that stays near zero means there is enough memory.
  • A chart has a gap. A gap is a period without samples (server restarted, unreachable or skipped under load). Gaps never change the state of a monitor.
  • The disk line is climbing and I do not know why. Look at logs, old releases and backups. See Disk full, out of memory or a slow server.
Last updated 2026-10-247 min readNeed help? Contact support
NextGet told when a scheduled task stops running