Monitoring and charts
Server overview and charts help track CPU, memory, disk and network activity. Start with current state and continue with historical charts.
Review usage
- Open the server details and charts.
- Select an available time range.
- Compare CPU, memory, disk and network values over the same period.
- Correlate peaks with deployment, backup or other operations in history.
A brief peak and sustained load need different responses. Identify the cause before changing resources.
History and downsampling
Selecting Hour / Day / Week / Month / Year changes both platform charts and stored guest measurements. Database charts show averages; the average/maximum selector applies only to platform charts.
Stored history uses these resolutions:
| Period | One point |
|---|---|
| Hour | 1 minute |
| Day | 30 minutes |
| Week | 3 hours |
| Month | 12 hours |
| Year | 7 days |
Each range contains 70 time intervals, matching platform history. Its actual duration may therefore differ from a calendar day, month or year. Missing measurements remain gaps rather than zeros or interpolated values. Charts show the last measurement time.
Raw samples are retained for one day. Aggregation continuously builds all history levels, including individual CPU cores and diagnostic metrics. As data ages, finer levels expire and coarser history remains; history outside the final yearly window is removed. Samples are deleted only after their aggregates have been saved successfully.
Existing hourly averages are imported as coarse fallback history. They cannot reconstruct missing detailed samples. Actual measurements take precedence when available for the same interval.
Interpret values
Allocated virtual server resources and values measured inside the operating system are not identical. For example, virtual disk size differs from filesystem free space.
Some details require an available guest agent or a prepared service. Missing data does not automatically mean zero usage. Information may be stale when the server is stopped or collection is interrupted.
Investigate a problem
- Memory: check process counts and application requirements.
- Disk: inspect filesystems, logs and application data.
- CPU: compare usage with running commands and traffic.
- Network: check actual service availability and the firewall.
Combine charts with logs and notifications. Also test web application availability through a real request.