max / makenotwork

4.5 KB · 130 lines History Blame Raw

1	//! Inbound infra-alert ingestion: PoM / MT monitoring agents → MNW.
2	//!
3	//! `POST /api/internal/alerts` accepts an ops alert (health failure, TLS
4	//! expiry, DNS change, ...) from an external monitoring agent, persists it to
5	//! `admin_alerts`, and emails the operator (`ALERT_EMAIL`). Admin-routed only:
6	//! these are operator signals, never creator-facing. Authed by [`AlertsAuth`], a
7	//! bearer token distinct from the CLI internal API's `ServiceAuth`.
8
9	use axum::{Json, extract::State, response::IntoResponse};
10	use serde::Deserialize;
11	use sqlx::PgPool;
12
13	use crate::{
14	auth::AlertsAuth,
15	db::{
16	self,
17	admin_alerts::{AlertKind, AlertSeverity},
18	},
19	email::EmailClient,
20	error::{AppError, Result},
21	};
22
23	#[derive(Deserialize)]
24	pub(super) struct AlertRequest {
25	/// Reporting agent, e.g. "pom" or "mt".
26	source: String,
27	/// Alert class. Unknown values are rejected at deserialization.
28	kind: AlertKind,
29	/// Alert urgency. Unknown values are rejected at deserialization.
30	severity: AlertSeverity,
31	/// Short human-readable headline.
32	title: String,
33	/// Full alert detail.
34	body: String,
35	/// Optional agent-side collapse key for repeated firings of one condition.
36	#[serde(default)]
37	dedup_key: Option<String>,
38	/// Optional structured context (JSON object), stored verbatim.
39	#[serde(default)]
40	details: Option<serde_json::Value>,
41	}
42
43	fn require_field(name: &str, value: &str, max: usize) -> Result<()> {
44	if value.is_empty() {
45	return Err(AppError::validation(format!("{name} is required")));
46	}
47	if value.len() > max {
48	return Err(AppError::validation(format!(
49	"{name} must be at most {max} characters"
50	)));
51	}
52	Ok(())
53	}
54
55	/// POST /api/internal/alerts
56	///
57	/// Persist an inbound alert and email the operator. Returns 202 with the row id.
58	/// A missing `ALERT_EMAIL` is not an error: the alert is still logged (the row's
59	/// `emailed` stays false so the gap is visible), because dropping the durable
60	/// record just because mail is unconfigured would defeat the point.
61	pub(super) async fn ingest_alert(
62	State(pool): State<PgPool>,
63	State(email): State<EmailClient>,
64	_auth: AlertsAuth,
65	Json(req): Json<AlertRequest>,
66	) -> Result<impl IntoResponse> {
67	// `kind` and `severity` are already validated by the type system: the
68	// request would not have deserialized if either were outside its enum. Only
69	// the free-text fields need bounds here.
70	require_field("source", &req.source, 64)?;
71	require_field("title", &req.title, 200)?;
72	require_field("body", &req.body, 5000)?;
73	if let Some(key) = &req.dedup_key
74	&& key.len() > 200
75	{
76	return Err(AppError::validation(
77	"dedup_key must be at most 200 characters".to_string(),
78	));
79	}
80
81	let id = db::admin_alerts::insert_alert(
82	&pool,
83	&db::admin_alerts::NewAlert {
84	source: &req.source,
85	kind: req.kind,
86	severity: req.severity,
87	title: &req.title,
88	body: &req.body,
89	dedup_key: req.dedup_key.as_deref(),
90	details: req.details.as_ref(),
91	},
92	)
93	.await?;
94
95	match std::env::var("ALERT_EMAIL").ok() {
96	Some(to) if !to.is_empty() => {
97	let subject = format!("[{}/{}] {}", req.source, req.severity.as_str(), req.title);
98	let mail_body = format!(
99	"{}\n\nsource: {}\nkind: {}\nseverity: {}\n",
100	req.body,
101	req.source,
102	req.kind.as_str(),
103	req.severity.as_str()
104	);
105	match email.send_alert(&to, &subject, &mail_body).await {
106	Ok(()) => {
107	if let Err(e) = db::admin_alerts::mark_emailed(&pool, id).await {
108	tracing::error!(alert_id = id, error = ?e, "alert emailed but mark_emailed failed");
109	}
110	}
111	// Ingestion succeeded and the row is durable; a mail failure is
112	// logged, not surfaced as a 5xx to the agent (which would make it
113	// retry and duplicate the row).
114	Err(e) => {
115	tracing::error!(alert_id = id, error = ?e, "failed to email ingested alert");
116	}
117	}
118	}
119	_ => tracing::warn!(
120	alert_id = id,
121	"ALERT_EMAIL unset; alert logged but not emailed"
122	),
123	}
124
125	Ok((
126	axum::http::StatusCode::ACCEPTED,
127	Json(serde_json::json!({ "id": id })),
128	))
129	}
130