Add circuit breaker for HA connectivity failures

When Home Assistant becomes unreachable (WiFi dropout, HA restart, etc.),
the module now:
- Detects connectivity errors (ENETUNREACH, timeout, etc.)
- Opens a circuit breaker that stops ALL template evaluations immediately
- Runs a single coalesced health check with exponential backoff (15s→300s)
- Queues pending evaluations and replays them when HA comes back
- Suppresses error log spam while circuit is open

Previously, each of the 11 sections would independently fail + retry,
flooding the event loop with 30+ error callbacks in the same millisecond
and freezing Electron's renderer.

Fixes repeated MagicMirror freezes during transient network outages.

node_helper.js

@@ -3,6 +3,12 @@
  *
  * By Brian Towles
  * MIT Licensed.
+ *
+ * Resilience improvements by James (2026-02-27):
+ * - Circuit breaker: stops all template evals when HA is unreachable
+ * - Exponential backoff on retries (no more 30 simultaneous retry timers)
+ * - Coalesced health checks instead of per-section retries
+ * - Graceful degradation: sections show stale data instead of freezing
  */
 var backoff = require('backoff')
 const NodeHelper = require("node_helper");
@@ -24,6 +30,10 @@ module.exports = NodeHelper.create({
 	evaluateTemplate,
 	onWebsocketCloseEvent,
 	backoffWSConnection,
+	_getBreaker,
+	_openCircuit,
+	_closeCircuit,
+	_healthCheck,
 });
 
 function start() {
@@ -32,6 +42,10 @@ function start() {
 		this.logger.debug("MMM-HomeAssistantDisplay helper started...");
 	}
 	this.connections = {};
+
+	// Circuit breaker state: tracks HA reachability per connection
+	// When open, template evaluations are skipped entirely (no requests, no error spam)
+	this._circuitBreaker = {};  // identifier -> { state: 'closed'|'open', failCount: 0, nextRetryAt: 0, retryTimer: null }
 }
 
 function stop() {
@@ -67,10 +81,11 @@ function socketNotificationReceived(notification, payload) {
 			this.evaluateTemplate(payload).then((ret) => {
 				this.sendSocketNotification("MODULE_DISPLAY_RENDERED", ret);
 			}).catch((err) => {
-				this.logger.error(
+				// Only log if circuit is closed (avoid spam when HA is down)
-					"Unable to evaluate template",
+				const cb = this._getBreaker(payload.identifier);
-					err
+				if (cb.state === 'closed') {
-				);
+					this.logger.error("Unable to evaluate template", err);
+				}
 			});
 			break;
 		case "RENDER_SECTION_DISPLAY_TEMPLATE":
@@ -80,19 +95,128 @@ function socketNotificationReceived(notification, payload) {
 					section: payload.section
 				});
 			}).catch((err) => {
-				this.logger.error(
+				const cb = this._getBreaker(payload.identifier);
-					"unable to evaluate section template",
+				if (cb.state === 'closed') {
-					err
+					this.logger.error("unable to evaluate section template", err);
-				);
+				}
 			});
 			break;			
 	}
 }
 
+// Circuit breaker helpers
+function _getBreaker(identifier) {
+	if (!this._circuitBreaker[identifier]) {
+		this._circuitBreaker[identifier] = {
+			state: 'closed',      // closed = healthy, open = HA unreachable
+			failCount: 0,
+			nextRetryAt: 0,
+			retryTimer: null,
+			pendingQueue: [],     // queued payloads to retry when circuit closes
+		};
+	}
+	return this._circuitBreaker[identifier];
+}
+
+function _openCircuit(identifier) {
+	const breaker = this._getBreaker(identifier);
+	if (breaker.state === 'open') return; // already open
+
+	breaker.state = 'open';
+	breaker.failCount++;
+	// Exponential backoff: 15s, 30s, 60s, 120s, max 300s
+	const delay = Math.min(15000 * Math.pow(2, breaker.failCount - 1), 300000);
+	breaker.nextRetryAt = Date.now() + delay;
+
+	this.logger.info(`Circuit OPEN for ${identifier} — HA unreachable. Next health check in ${delay / 1000}s`);
+
+	// Single coalesced health check timer (not per-section!)
+	if (breaker.retryTimer) clearTimeout(breaker.retryTimer);
+	breaker.retryTimer = setTimeout(() => {
+		this._healthCheck(identifier);
+	}, delay);
+}
+
+function _closeCircuit(identifier) {
+	const breaker = this._getBreaker(identifier);
+	if (breaker.state === 'closed') return;
+
+	this.logger.info(`Circuit CLOSED for ${identifier} — HA is reachable again`);
+	breaker.state = 'closed';
+	breaker.failCount = 0;
+	breaker.nextRetryAt = 0;
+	if (breaker.retryTimer) {
+		clearTimeout(breaker.retryTimer);
+		breaker.retryTimer = null;
+	}
+
+	// Replay queued template evaluations
+	const queue = breaker.pendingQueue.splice(0);
+	if (queue.length > 0) {
+		this.logger.info(`Replaying ${queue.length} queued template evaluations for ${identifier}`);
+		for (const item of queue) {
+			this.evaluateTemplate(item.payload).then((ret) => {
+				if (item.payload.section !== undefined) {
+					this.sendSocketNotification("SECTION_DISPLAY_RENDERED", {
+						...ret,
+						section: item.payload.section
+					});
+				} else {
+					this.sendSocketNotification("MODULE_DISPLAY_RENDERED", ret);
+				}
+			}).catch(() => {
+				// Circuit will re-open if this fails
+			});
+		}
+	}
+}
+
+async function _healthCheck(identifier) {
+	const breaker = this._getBreaker(identifier);
+	const hass = this.connections[identifier] && this.connections[identifier].hass;
+	if (!hass) return;
+
+	this.logger.info(`Health check for ${identifier}...`);
+	try {
+		// Simple template render as health probe
+		await Promise.race([
+			hass.templates.render("{{ 1 }}"),
+			new Promise((_, reject) =>
+				setTimeout(() => reject(new Error('Health check timeout')), 10000)
+			)
+		]);
+		// Success — close the circuit
+		this._closeCircuit(identifier);
+	} catch (err) {
+		this.logger.info(`Health check failed for ${identifier}: ${err.message}`);
+		// Re-arm with increased backoff
+		breaker.state = 'closed'; // briefly close so _openCircuit fires
+		this._openCircuit(identifier);
+	}
+}
+
 async function evaluateTemplate(payload) {
 	if (config.debuglogging) {
 		this.logger.debug(`Evaluating template for ${payload.template}`);
 	}
+
+	// Circuit breaker check: if HA is known-unreachable, skip entirely
+	const breaker = this._getBreaker(payload.identifier);
+	if (breaker.state === 'open') {
+		// Queue this request for replay when circuit closes (deduplicated by section)
+		const isDuplicate = breaker.pendingQueue.some(
+			(item) => item.payload.section === payload.section && item.payload.template === payload.template
+		);
+		if (!isDuplicate) {
+			breaker.pendingQueue.push({ payload });
+			// Cap queue size to prevent memory growth
+			if (breaker.pendingQueue.length > 20) {
+				breaker.pendingQueue.shift();
+			}
+		}
+		throw new Error('Circuit open — HA unreachable, skipping template evaluation');
+	}
+
 	const hass = this.connections[payload.identifier].hass;
 
 	try {
@@ -103,32 +227,28 @@ async function evaluateTemplate(payload) {
 				setTimeout(() => reject(new Error('Template evaluation timeout')), 10000)
 			)
 		]);
+
+		// Success — ensure circuit is closed (resets fail count)
+		if (breaker.failCount > 0) {
+			this._closeCircuit(payload.identifier);
+		}
+
 		return {
 			identifier: payload.identifier,
 			render: response
 		};
 	} catch (err) {
-		this.logger.error(`Template evaluation failed: ${err.message}`);
+		// Check if this is a connectivity error (not a template syntax error)
+		const isConnectivityError = /ENETUNREACH|ECONNREFUSED|ECONNRESET|ETIMEDOUT|EHOSTUNREACH|timeout/i.test(err.message);
 
-		// Schedule retry after 30 seconds
+		if (isConnectivityError) {
-		setTimeout(() => {
+			this.logger.error(`Template evaluation failed (connectivity): ${err.message}`);
-			this.logger.info(`Retrying template evaluation for ${payload.identifier}`);
+			this._openCircuit(payload.identifier);
-			this.evaluateTemplate(payload).then((ret) => {
+		} else {
-				// Send appropriate notification based on original request type
+			// Template/logic error — log but don't trip the breaker
-				if (payload.section !== undefined) {
+			this.logger.error(`Template evaluation failed: ${err.message}`);
-					this.sendSocketNotification("SECTION_DISPLAY_RENDERED", {
+		}
-						...ret,
-						section: payload.section
-					});
-				} else {
-					this.sendSocketNotification("MODULE_DISPLAY_RENDERED", ret);
-				}
-			}).catch((retryErr) => {
-				this.logger.error(`Template evaluation retry also failed: ${retryErr.message}`);
-			});
-		}, 30000);
 
-		// Re-throw to maintain existing error handling behavior
 		throw err;
 	}
 }