What the charts show
The panel stores sanitized periodic samples from the infrastructure provider. Short gaps can occur when the provider or VPS is unreachable. Charts are useful for trends, not a substitute for guest-level application monitoring.
Interpret carefully
A brief CPU spike is different from sustained saturation. High memory usage may include reclaimable cache. Disk capacity and disk I/O are different measurements. Network totals do not identify which process generated traffic.
Investigate
Correlate the chart time with deployments, backups, traffic, and application logs. Use a reviewed health report or read-only guest tools for confirmation. Increase resources only after identifying the bottleneck and keeping a current backup.