feat(host-agent): Rust rewrite Phase 0 — multi-instance foundation, v2 wire protocol, real telemetry
All checks were successful
Test Asgard Runner / test (push) Successful in 3s
All checks were successful
Test Asgard Runner / test (push) Successful in 3s
New corrosion-host-agent/ crate (Go companion-agent stays as behavior
reference until parity). Wire protocol v2 per COA-B: instance-scoped
subjects corrosion.{license}.{instance}.* + host-level .host.* — spec
in PROTOCOL.md, designed for the license->host->instance fleet model.
- Multi-instance TOML config in the foundation, not retrofitted
- NATS layer on the Vigilance production profile (infinite reconnect,
capped backoff, 30s ping, 8192-msg offline buffer)
- Heartbeat with real sysinfo telemetry — Go agent shipped hardcoded
disk/cpu placeholders; this is the panel's first true Resources data
- Connectivity prober (outbound TCP, periodic + on-demand)
- Host cmd channel (ping/probe/sysinfo), going-offline beacon,
CancellationToken shutdown
- Live-fire verified against production NATS; artifacts: 3.7MB static
linux-musl, 3.8MB windows .exe (static CRT)
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
115
corrosion-host-agent/src/hostcmd.rs
Normal file
115
corrosion-host-agent/src/hostcmd.rs
Normal file
@@ -0,0 +1,115 @@
|
||||
//! Host-level command handler: request-reply on `corrosion.{license}.host.cmd`.
|
||||
//!
|
||||
//! One subscriber; each message handled in its own task so a slow command
|
||||
//! never blocks the dispatch loop. Phase 0 commands: ping, probe, sysinfo.
|
||||
|
||||
use futures::StreamExt;
|
||||
use serde::Deserialize;
|
||||
use serde_json::json;
|
||||
use std::sync::Arc;
|
||||
use sysinfo::System;
|
||||
|
||||
use crate::agent::Agent;
|
||||
use crate::prober;
|
||||
use crate::subjects;
|
||||
use crate::telemetry;
|
||||
use crate::version;
|
||||
|
||||
#[derive(Debug, Deserialize)]
|
||||
struct HostCommand {
|
||||
func: String,
|
||||
}
|
||||
|
||||
pub async fn run(agent: Arc<Agent>) -> anyhow::Result<()> {
|
||||
let subject = subjects::host_cmd(&agent.cfg.license_id);
|
||||
let mut sub = agent.nats.subscribe(subject.clone()).await?;
|
||||
tracing::info!("host command handler listening on {subject}");
|
||||
|
||||
let cancel = agent.shutdown.clone();
|
||||
loop {
|
||||
tokio::select! {
|
||||
msg = sub.next() => {
|
||||
match msg {
|
||||
Some(msg) => {
|
||||
let agent = agent.clone();
|
||||
tokio::spawn(async move { handle(agent, msg).await });
|
||||
}
|
||||
None => {
|
||||
tracing::warn!("host command subscription ended");
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
_ = cancel.cancelled() => {
|
||||
tracing::info!("host command handler stopping");
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
Ok(())
|
||||
}
|
||||
|
||||
async fn handle(agent: Arc<Agent>, msg: async_nats::Message) {
|
||||
let Some(reply) = msg.reply.clone() else {
|
||||
tracing::warn!("host command without reply subject ignored");
|
||||
return;
|
||||
};
|
||||
|
||||
let response = match serde_json::from_slice::<HostCommand>(&msg.payload) {
|
||||
Ok(cmd) => dispatch(&agent, &cmd.func).await,
|
||||
Err(e) => json!({ "status": "error", "message": format!("invalid command payload: {e}") }),
|
||||
};
|
||||
|
||||
let bytes = match serde_json::to_vec(&response) {
|
||||
Ok(b) => b,
|
||||
Err(e) => {
|
||||
tracing::error!("response serialize failed: {e}");
|
||||
return;
|
||||
}
|
||||
};
|
||||
if let Err(e) = agent.nats.publish(reply, bytes.into()).await {
|
||||
tracing::warn!("response publish failed: {e}");
|
||||
}
|
||||
}
|
||||
|
||||
async fn dispatch(agent: &Arc<Agent>, func: &str) -> serde_json::Value {
|
||||
match func {
|
||||
"ping" => json!({
|
||||
"status": "success",
|
||||
"func": "ping",
|
||||
"version": version::VERSION,
|
||||
"commit": version::GIT_HASH,
|
||||
"uptime_seconds": agent.started.elapsed().as_secs(),
|
||||
}),
|
||||
"probe" => {
|
||||
let report = prober::run_probe(&agent.cfg.probe_targets).await;
|
||||
*agent.last_probe.write().await = Some(report.clone());
|
||||
match serde_json::to_value(&report) {
|
||||
Ok(report_json) => json!({
|
||||
"status": "success",
|
||||
"func": "probe",
|
||||
"report": report_json,
|
||||
}),
|
||||
Err(e) => json!({ "status": "error", "message": format!("probe serialize: {e}") }),
|
||||
}
|
||||
}
|
||||
"sysinfo" => {
|
||||
let mut sys = System::new();
|
||||
sys.refresh_cpu_usage();
|
||||
tokio::time::sleep(std::time::Duration::from_millis(250)).await;
|
||||
let payload = telemetry::collect(agent, &mut sys).await;
|
||||
match serde_json::to_value(&payload) {
|
||||
Ok(snapshot) => json!({
|
||||
"status": "success",
|
||||
"func": "sysinfo",
|
||||
"snapshot": snapshot,
|
||||
}),
|
||||
Err(e) => json!({ "status": "error", "message": format!("sysinfo serialize: {e}") }),
|
||||
}
|
||||
}
|
||||
other => json!({
|
||||
"status": "error",
|
||||
"message": format!("unknown func '{other}' (supported: ping, probe, sysinfo)"),
|
||||
}),
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user