Christoffer Martinsson
2581435b10
Replaced system-wide disk usage with accurate per-service tracking by scanning service-specific directories. Services like sshd now correctly show minimal disk usage instead of misleading system totals. - Rename storage widget and add drive capacity/usage columns - Move host display to main dashboard title for cleaner layout - Replace separate alert displays with color-coded row highlighting - Add per-service disk usage collection using du command - Update services widget formatting to handle small disk values - Restructure into workspace with dedicated agent and dashboard packages
149 lines
6.3 KiB
Markdown
149 lines
6.3 KiB
Markdown
# CM Dashboard
|
||
|
||
CM Dashboard is a Rust-powered terminal UI for real-time monitoring of CMTEC infrastructure hosts. It subscribes to the CMTEC ZMQ gossip network where lightweight agents publish SMART, service, and backup metrics, and presents them in an efficient, keyboard-driven interface built with `ratatui`.
|
||
|
||
```
|
||
┌──────────────────────────────────────────────────────────────────────────────┐
|
||
│ CM Dashboard │
|
||
├────────────────────────────┬────────────────────────────┬────────────────────┤
|
||
│ NVMe Health │ Services │ CPU / Memory │
|
||
│ Host: srv01 │ Host: srv01 │ Host: srv01 │
|
||
│ Status: Healthy │ Service memory: 1.2G/4.0G │ RAM: 6.9 / 7.8 GiB │
|
||
│ Healthy/Warning/Critical: │ Disk usage: 45 / 500 GiB │ CPU load (1/5/15): │
|
||
│ 4 / 0 / 0 │ Services tracked: 8 │ 1.2 0.9 0.7 │
|
||
│ Capacity used: 512 / 2048G │ │ CPU temp: 68°C │
|
||
│ Issue: — │ nginx running 320M │ GPU temp: — │
|
||
│ │ immich running 1.2G │ Status • ok │
|
||
│ │ backup-api running 40M │ │
|
||
├────────────────────────────┴────────────┬───────────────┴────────────────────┤
|
||
│ Backups │ Alerts │
|
||
│ Host: srv01 │ srv01: ok │
|
||
│ Overall: Healthy │ labbox: warning: RAM 82% │
|
||
│ Last success: 2024-02-01 03:12:45 │ cmbox: critical: CPU temp 92°C │
|
||
│ Snapshots: 17 • Size: 512.0 GiB │ Update: 2024-02-01 10:15:32 │
|
||
│ Pending jobs: 0 (enabled: true) │ │
|
||
└──────────────────────────────┬───────────────────────────────────────────────┘
|
||
│ Status │ │
|
||
│ Active host: srv01 (1/3) │ History retention ≈ 3600s │
|
||
│ Config: config/dashboard.toml│ Default host: labbox │
|
||
└──────────────────────────────┴───────────────────────────────────────────────┘
|
||
```
|
||
|
||
## Requirements
|
||
|
||
- Rust toolchain 1.75+ (install via [`rustup`](https://rustup.rs))
|
||
- Network access to the CMTEC metrics gossip agents (default `tcp://<host>:6130`; install `zeromq`/`libzmq` on the host)
|
||
- Configuration files under `config/` describing hosts and dashboard preferences
|
||
|
||
## Installation
|
||
|
||
Clone the repository and build with Cargo:
|
||
|
||
```bash
|
||
git clone https://github.com/cmtec/cm-dashboard.git
|
||
cd cm-dashboard
|
||
cargo build --release
|
||
```
|
||
|
||
The optimized binary is available at `target/release/cm-dashboard`. To install into your Cargo bin directory:
|
||
|
||
```bash
|
||
cargo install --path dashboard
|
||
```
|
||
|
||
## Configuration
|
||
|
||
On first launch, the dashboard will create `config/dashboard.toml` and `config/hosts.toml` automatically if they do not exist.
|
||
|
||
You can also generate starter configuration files manually with the built-in helper:
|
||
|
||
```bash
|
||
cargo run -p cm-dashboard -- init-config
|
||
# or, once installed
|
||
cm-dashboard init-config --dir ./config --force
|
||
```
|
||
|
||
This produces `config/dashboard.toml` and `config/hosts.toml`. The primary dashboard config looks like:
|
||
|
||
```toml
|
||
[hosts]
|
||
default_host = "srv01"
|
||
|
||
[[hosts.hosts]]
|
||
name = "srv01"
|
||
enabled = true
|
||
|
||
[[hosts.hosts]]
|
||
name = "labbox"
|
||
enabled = true
|
||
|
||
[dashboard]
|
||
tick_rate_ms = 250
|
||
history_duration_minutes = 60
|
||
|
||
[[dashboard.widgets]]
|
||
id = "nvme"
|
||
enabled = true
|
||
|
||
[[dashboard.widgets]]
|
||
id = "alerts"
|
||
enabled = true
|
||
|
||
[data_source]
|
||
kind = "zmq"
|
||
|
||
[data_source.zmq]
|
||
endpoints = ["tcp://127.0.0.1:6130"]
|
||
```
|
||
|
||
Adjust the host list and `data_source.zmq.endpoints` to match your CMTEC gossip network. If you prefer to manage hosts separately, edit the generated `hosts.toml` file.
|
||
|
||
## Features
|
||
|
||
- Rotating host selection with left/right arrows (`←`, `→`, `h`, `l`, `Tab`)
|
||
- Live NVMe, service, CPU/memory, backup, and alert panels per host
|
||
- Health scoring that rolls CPU/RAM/GPU pressure into alerts automatically
|
||
- Structured logging with `tracing` (`-v`/`-vv` to increase verbosity)
|
||
- Help overlay (`?`) outlining keyboard shortcuts
|
||
- Config-driven host discovery via `config/dashboard.toml`
|
||
|
||
## Getting Started
|
||
|
||
```bash
|
||
cargo run -p cm-dashboard -- --config config/dashboard.toml
|
||
# specify a single host
|
||
cargo run -p cm-dashboard -- --host srv01
|
||
# override ZMQ endpoints at runtime
|
||
cargo run -p cm-dashboard -- --zmq-endpoint tcp://srv01:6130,tcp://labbox:6130
|
||
# increase logging verbosity
|
||
cargo run -p cm-dashboard -- -v
|
||
```
|
||
|
||
### Keyboard Shortcuts
|
||
|
||
| Key | Action |
|
||
| --- | --- |
|
||
| `←` / `h` | Previous host |
|
||
| `→` / `l` / `Tab` | Next host |
|
||
| `?` | Toggle help overlay |
|
||
| `r` | Update status message |
|
||
| `q` / `Esc` | Quit |
|
||
|
||
## Agent
|
||
|
||
The metrics agent publishes SMART/service/backup data to the gossip network. Run it on each host (or under systemd/NixOS) and point the dashboard at its endpoint. Example:
|
||
|
||
```bash
|
||
cargo run -p cm-dashboard-agent -- --hostname srv01 --bind tcp://*:6130 --interval-ms 5000
|
||
```
|
||
|
||
Use `--disable-*` flags to skip collectors when a host doesn’t expose those metrics.
|
||
|
||
## Development
|
||
|
||
- Format: `cargo fmt`
|
||
- Check workspace: `cargo check`
|
||
- Build release binaries: `cargo build --release`
|
||
|
||
The dashboard subscribes to the CMTEC ZMQ gossip network (default `tcp://127.0.0.1:6130`). Received metrics are cached per host and retained in an in-memory ring buffer for future trend analysis.
|