Remote Host Monitoring

NyaTerm provides five right-side monitoring panels for SSH sessions: Resource Monitor, NVIDIA GPU Monitor, Ascend NPU Monitor, Process Manager, and Docker Manager. They share some behavior:

• They only make sense for an SSH session, and bind only to a genuinely active SSH session
• Each is shown or hidden by its own toggle in Settings → Terminal; turning a toggle off also hides its activity-bar icon
• GPU, NPU, process, and Docker panels accept 3 to 120 seconds; the resource-monitor panel separately accepts 1 to 60 seconds
• A panel stops refreshing after several consecutive polling failures, to avoid repeatedly hitting an unsupported host

Default toggle states are below. Note that Process Manager and Docker Manager are on by default:

Panel	Setting	Default	Default interval
Resource Monitor	Show remote resource info	On	3 s
NVIDIA GPU Monitor	Show NVIDIA GPU Monitor	Off	3 s
Ascend NPU Monitor	Show Ascend NPU Monitor	Off	3 s
Process Manager	Show Process Manager	On	5 s
Docker Manager	Show Docker Manager	On	10 s

Resource Monitor

Remote resource monitoring is on by default. To see data, both of these must be true:

1. The current tab is an SSH session
2. Show Remote Resource Stats is enabled in Settings → Terminal

When enabled, the Resource Monitor icon appears in the right activity bar and the panel polls the host on the configured interval. The default interval is 3 seconds, and you can change it manually.

The panel displays:

• Hostname, OS, architecture, uptime
• Load average
• CPU usage
• Memory usage
• Network throughput

NVIDIA GPU Monitor

The NVIDIA GPU Monitor panel shows NVIDIA GPU status on the remote host. It is off by default; enable Show NVIDIA GPU Monitor in Settings → Terminal.

The panel displays:

• Driver version and CUDA version
• Summary: GPU count, highest utilization, memory usage, highest temperature
• Per-GPU card: index, model, performance state (pstate), utilization and memory bars; expand for UUID, temperature, power draw, fan speed, and free memory. Utilization above 70% / 90% is color-coded differently
• A searchable GPU process list (filter by PID, GPU index, user, or process name), sorted by GPU memory used

If the remote host has no NVIDIA GPU or is missing nvidia-smi, the panel shows a matching empty state.

Ascend NPU Monitor

The Ascend NPU Monitor panel shows Ascend NPU status on the remote host. It is off by default; enable Show Ascend NPU Monitor in Settings → Terminal.

The panel displays:

• Driver version and CANN version
• Summary: NPU count, highest AI Core utilization, memory usage, highest temperature
• Per device: device index, Physical ID, Bus ID, AI Core utilization, memory usage, temperature, and power draw
• A searchable NPU compute process list

If the remote host returns no Ascend NPU information, the panel shows a matching empty state.

Process Manager

The Process Manager panel shows a live process list from the remote host. It is on by default; turn off Show Process Manager in Settings → Terminal to hide it.

Key capabilities:

• Total process count and a search box (filter by PID, user, state, command, or full command line)
• Adaptive layout that adds or drops columns based on panel width; sort by process name, PID, CPU%, MEM%, or user
• Expand a process for PID/PPID, user, state, CPU%, memory%, RSS, elapsed time, and the full command line, plus adjusting the nice value (-20 to 19) and clicking Apply (renice)
• A row action menu to copy the PID or command, or send TERM / HUP / STOP / CONT signals; KILL first opens a confirmation dialog showing the kill command

If the remote host does not support process queries, the panel shows a distinct message.

Docker Manager

The Docker Manager panel manages Docker on the remote host. It is on by default; turn off Show Docker Manager in Settings → Terminal to hide it.

Key capabilities:

• Overview: running / stopped container counts and image count, with the Docker engine version in the header
• Global search plus tabs for containers, images, volumes, networks, and Compose (when available); extra tabs collapse into a More dropdown
• Containers: a state-sorted virtualized list; a row menu views logs (runs docker logs -f in the terminal), enters the container (opens a shell), starts / stops / restarts / kills (confirm) / removes (confirm); clicking a row opens a live-refreshing details dialog
• Images / Volumes / Networks: fetched on demand, each row supports removal (confirm)
• Compose: lists projects; expand to lazily load services; supports project-level up / restart / down and service-level logs / enter / up / stop / restart
• The More menu offers docker system prune (destructive, confirmed)

Logs and enter-container actions run in the real terminal session; remove, kill, Compose down, prune, and other destructive operations route through a confirmation dialog showing the exact command.