Track CPU, memory, disk, and network in real time - with thresholds that alert you before things break.
Every linked server continuously reports CPU, memory, swap, disk space and disk I/O utilization, network throughput, system load, uptime, and process counts. Per-mount and per-interface breakdowns mean you see the 42 GB / path filling up before the full-disk alert fires, and you can tell which network interface is saturated without guessing. All data is yours, tied to your organization, and retained according to your plan.
Every metric can be viewed as a smooth line, step area, gradient fill, multi-line comparison, or bar chart - whichever shows the pattern most clearly. Time windows range from the last 5 minutes to the last 30 days, and you can drag directly on any chart to zoom into a specific incident window. Charts are theme-aware and respond instantly when you switch between dark and light modes.
Thresholds are configured per project under the Monitoring settings page. You can combine percentage rules (CPU above 90%, memory above 85%, disk above 80%) with absolute ones (less than 2 GB of free disk, less than 512 MB of free RAM) so you get meaningful alerts on both tiny and huge machines. Disk I/O utilization, load average, and swap pressure each have their own dedicated thresholds.
Dedicated and bare-metal servers expose physical sensors that virtual machines do not, and Optimux surfaces them in a dedicated Hardware tab. You get CPU and disk temperatures with history, per-disk health and wear (SMART) including drive age and total data written, memory (ECC) error counts, RAID array status with rebuild progress, fan speeds, and power and baseboard (IPMI) sensors. Set warning and critical temperature thresholds per server, and Optimux alerts you on overheating, a worn or failing disk, a stalled fan, a power or voltage problem, a PCIe link error, or a filesystem forced read-only. Virtual machines show their type, virtual CPU, memory, and disks instead.
The platform tracks when each server last reported in. If a server goes silent for longer than your configured timeout (anywhere from 1 to 60 minutes), it is flagged as offline and an alert fires so you notice before customers do. When you deliberately put a server into Maintenance mode, for an OS upgrade, for example, health checks, scans, and alerts are paused automatically so you are not paged for something you already know about.
The Overview page pulls the most important signals into one screen: active alerts, top servers by CPU and memory, scan status, uptime summary, and recent activity. You can drag the cards into the order that matches how you work, and the layout is remembered per user. Clicking any widget jumps straight to the underlying detail page, so triage takes two clicks rather than ten.
