fix(notifications): resolve version notification showing 0.0.0 and backfill missing image update notifications (#586)

- Read running Sencho version from the packaged manifest via getSenchoVersion() instead of process.env.npm_package_version, which is undefined when launched via node dist/index.js (production Docker). Skip the update check entirely when the version cannot be resolved.
- Add a one-time backfill pass in ImageUpdateService so users who upgraded to a Sencho version with the notification pipeline receive a catch-up entry for stacks already flagged as having updates before the upgrade.
- Surface dispatch failures as error-level entries in the in-app notification bell via a direct notification_history write, so misconfigured webhooks are visible without tailing logs.
- Extend test coverage for both paths: mock getSenchoVersion (including the null-version case) and add dispatch-path tests for transition, no-re-fire, backfill, and error surfacing.
- Expand alerts-notifications docs with per-instance 6-hour check cadence, an example version message, a backfill note, and three new troubleshooting entries.

backend/src/services/ImageUpdateService.ts

@@ -9,6 +9,9 @@ import { RegistryService } from './RegistryService';
 import { NodeRegistry } from './NodeRegistry';
 import { NotificationService } from './NotificationService';
 import { isDebugEnabled } from '../utils/debug';
+import { getErrorMessage } from '../utils/errors';
+
+const BACKFILL_KEY = 'image_update_notifications_backfilled';
 
 // ─── Image ref parsing ────────────────────────────────────────────────────────
 
@@ -377,6 +380,10 @@ export class ImageUpdateService {
         // Read previous state to detect new updates for notifications
         const previousState = db.getStackUpdateStatus(nodeId);
 
+        // One-time backfill: pre-existing has_update rows predate the notification pipeline;
+        // treat them as unnotified on first run so users get a catch-up entry per affected stack.
+        const isBackfilled = db.getSystemState(BACKFILL_KEY) === '1';
+
         // Write status for ALL stacks (including those with no pullable images)
         const now = Date.now();
         let updatesFound = 0;
@@ -386,7 +393,7 @@ export class ImageUpdateService {
             if (hasUpdate) {
                 updatesFound++;
                 // Notify only on state transition: was false/absent, now true
-                if (!previousState[stackName]) {
+                if (!isBackfilled || !previousState[stackName]) {
                     newlyUpdated.push(stackName);
                 }
             }
@@ -405,10 +412,25 @@ export class ImageUpdateService {
                     );
                 } catch (e) {
                     console.error(`[ImageUpdateService] Failed to dispatch update notification for "${stackName}":`, e);
+                    // Direct DB write to avoid recursing through dispatchAlert if it is what failed.
+                    try {
+                        db.addNotificationHistory({
+                            level: 'error',
+                            message: `[Node: ${nodeName}] Failed to notify about image updates for stack "${stackName}": ${getErrorMessage(e, String(e))}`,
+                            timestamp: Date.now(),
+                        });
+                    } catch (dbErr) {
+                        console.error('[ImageUpdateService] Failed to record dispatch error:', dbErr);
+                    }
                 }
             }
         }
 
+        // Mark the backfill flag after the first run so future checks use strict transitions.
+        if (!isBackfilled) {
+            db.setSystemState(BACKFILL_KEY, '1');
+        }
+
         console.log(`[ImageUpdateService] Node ${nodeId}: checked ${allImages.size} image(s), ${updatesFound} stack(s) with updates`);
 
         // Prune stale entries for stacks no longer on disk (reuse previousState to avoid extra DB read)

backend/src/services/MonitorService.ts

@@ -5,7 +5,7 @@ import semver from 'semver';
 import DockerController from './DockerController';
 import { DatabaseService } from './DatabaseService';
 import { NotificationService } from './NotificationService';
-import { isValidVersion } from './CapabilityRegistry';
+import { isValidVersion, getSenchoVersion } from './CapabilityRegistry';
 import { fetchLatestSenchoVersion } from '../utils/version-check';
 import { isDebugEnabled } from '../utils/debug';
 
@@ -266,7 +266,9 @@ export class MonitorService {
         if (Date.now() - this.lastVersionCheckAt > MonitorService.VERSION_CHECK_INTERVAL_MS) {
             this.lastVersionCheckAt = Date.now();
             try {
-                const currentVersion = process.env.npm_package_version || '0.0.0';
+                // Resolve from the packaged manifest: process.env.npm_package_version is
+                // only set by npm scripts, so it is undefined in Docker (node dist/index.js).
+                const currentVersion = getSenchoVersion();
                 const latest = await fetchLatestSenchoVersion();
                 if (isValidVersion(latest) && isValidVersion(currentVersion) && semver.gt(latest, currentVersion)) {
                     const db = DatabaseService.getInstance();
@@ -278,6 +280,8 @@ export class MonitorService {
                             `Sencho ${latest} is available (currently running ${currentVersion}). Visit the Fleet dashboard to update.`);
                         db.setSystemState(stateKey, latest);
                     }
+                } else if (isDebugEnabled() && !isValidVersion(currentVersion)) {
+                    console.debug('[Monitor:diag] Sencho version unresolvable; skipping update notification');
                 }
             } catch (e) {
                 // Network errors are expected; do not spam logs