System Monitor

System Monitor combines CPU, auto-detected GPU, memory, storage, and network metrics into one compact Noctalia bar capsule. Subtle dividers separate non-empty groups, and values can use the global Noctalia system-monitor critical threshold for highlighting.

Plugin

Field	Value
ID	tmelik/system-monitor
Entries	Bar widget: summary
License	GPL-3.0-only

Requirements

• ps from procps/procps-ng for the optional top-process tooltip rows.
• cat (coreutils) for the CPU-frequency fallback used only when Noctalia doesn't report it.
• Noctalia v5 with plugin API 26 or newer.
• The Noctalia system monitor service must be enabled.

Usage

1. Open Noctalia Settings.
2. Go to Bar, add a plugin widget, and select System Monitor.
3. Place the widget in the desired section of the bar.

Left-click the capsule to open the System tab in Control Center. Hover it to see a two-column details table. Rows follow the capsule's CPU, GPU, memory, storage, and network order; CPU usage, temperature, frequency, and load are opt-in via show_cpu_details_in_tooltip since the capsule already shows enabled CPU metrics. RAM, swap, disk, and VRAM rows show utilization plus used / total amounts. The network row shows RX/TX accumulated during the current plugin session instead of repeating the live rates from the capsule. The final rows identify the highest CPU and RAM consumers.

Middle-click the capsule to open its settings. Each metric can be shown or hidden independently; GPU metrics default to auto (shown only when a GPU reports that field). You can also toggle category separators and the detailed tooltip, choose compact or explicit network units, select a network interface, choose percentage/used/available displays for RAM and disk, and change the monitored disk path. The default path is the root filesystem (/).

Unavailable sensors are displayed as an em dash instead of a misleading zero, or can be hidden. An unknown network interface is unavailable and never silently falls back to total traffic. RAM uses a distinct server-module glyph so it is easy to tell apart from CPU at a glance.

Settings

Setting	Type	Default	Description
show_cpu_usage	bool	true	Show aggregate CPU usage.
show_cpu_temperature	bool	true	Show CPU temperature when a sensor is available.
show_gpu_usage	select	auto	auto/on/off. Auto shows GPU utilization when a GPU reports it.
show_gpu_temperature	select	auto	auto/on/off. Auto shows GPU temperature when a GPU reports it.
show_gpu_vram	select	auto	auto/on/off. Auto shows GPU VRAM usage when a GPU reports it.
show_ram	bool	true	Show RAM utilization.
show_swap	bool	true	Show swap utilization.
show_disk	bool	true	Show utilization for disk_path.
show_download	bool	true	Show aggregate receive rate.
show_upload	bool	true	Show aggregate transmit rate.
highlight_values	bool	true	Color glyphs and labels at Noctalia's activity and critical thresholds.
activity_color	color	primary	Color used from the activity threshold.
critical_color	color	error	Color used from the critical threshold.
hide_unavailable	bool	false	Omit individual metrics whose current value is unavailable.
show_separators	bool	true	Separate non-empty CPU, GPU, memory, storage, and network groups.
show_tooltip	bool	true	Show expanded metric names and values on hover.
show_cpu_details_in_tooltip	bool	false	Show CPU usage, temperature, frequency, and load in the tooltip.
show_top_processes	bool	true	Show the top CPU and RAM processes in the tooltip.
compact_network	bool	true	Use M/s-style labels; disable for MB/s-style labels.
network_interface	string	""	Exact interface name; empty uses total non-loopback traffic.
ram_display_mode	select	percentage	Show RAM as percentage, used, or available.
disk_display_mode	select	percentage	Show disk as percentage, used, or available.
disk_path	string	/	Select the filesystem whose utilization is displayed.

Value highlighting

When highlighting is enabled, both the glyph and label use the normal widget color below the activity threshold, activity_color from activity up to critical, and critical_color at critical or above. Unavailable values are never colored. RAM and disk compare their usage percentage even in absolute display modes; network rates compare decimal MB/s.

The widget reads the effective system.monitor.*_activity_threshold and *_critical_threshold values on every update. Missing or invalid pairs use Noctalia's defaults: CPU usage 50/90, CPU temperature 60/85, GPU usage 50/95, GPU temperature 60/85, VRAM 50/90, RAM 60/90, swap 20/80, disk 80/95, and RX/TX 1/50 MB/s. A global critical threshold of zero disables highlighting for that metric, matching Noctalia's system-monitor behavior.

Notes

• The plugin only reads snapshots exposed by Noctalia through systemStats() and diskStats(), plus the global system.monitor critical threshold settings.
• It does not access the network or write data.
• When show_top_processes and the tooltip are enabled, it runs ps asynchronously every five seconds. Only compact PID, short process name (comm), RSS, and accumulated CPU-time fields are requested; usernames, command lines, arguments, and environment variables are not read or displayed.
• When show_cpu_details_in_tooltip is enabled and Noctalia doesn't report CPU frequency itself, it falls back to reading /sys/devices/system/cpu/cpu0/cpufreq/scaling_cur_freq, then /proc/cpuinfo if that's unavailable, every five seconds.
• Values refresh according to Noctalia's system-monitor sampling intervals; the widget redraws once per second.
• The tooltip includes only metrics that are both enabled and rendered. Its amount rows do not include available memory or sample age.
• Network session totals are estimated from Noctalia's sampled rates. They start at zero when the plugin runtime loads, do not persist across a Noctalia reload/restart, and are not system-uptime counters. An explicitly selected interface has its own session totals.
• If every metric is disabled, the capsule shows No metrics. If enabled metrics are all hidden because they are unavailable, it shows No data.

Development

Repository knowledge and the current handoff follow Google Open Knowledge Format v0.2. Start at knowledge/index.md.

Run the repository-local check before publishing:

python3 scripts/validate_okf.py