agent resilliency

src/Agent/AutopilotMonitor.Agent.Core/Configuration/RemoteConfigService.cs

@@ -19,6 +19,7 @@ namespace AutopilotMonitor.Agent.Core.Configuration
         private readonly string _tenantId;
         private readonly AgentLogger _logger;
         private readonly string _cacheFilePath;
+        private readonly EmergencyReporter _emergencyReporter;
 
         private AgentConfigResponse _currentConfig;
         private readonly object _configLock = new object();
@@ -37,11 +38,12 @@ namespace AutopilotMonitor.Agent.Core.Configuration
             }
         }
 
-        public RemoteConfigService(BackendApiClient apiClient, string tenantId, AgentLogger logger)
+        public RemoteConfigService(BackendApiClient apiClient, string tenantId, AgentLogger logger, EmergencyReporter emergencyReporter = null)
         {
             _apiClient = apiClient ?? throw new ArgumentNullException(nameof(apiClient));
             _tenantId = tenantId ?? throw new ArgumentNullException(nameof(tenantId));
             _logger = logger ?? throw new ArgumentNullException(nameof(logger));
+            _emergencyReporter = emergencyReporter;
 
             var cacheDir = Environment.ExpandEnvironmentVariables(@"%ProgramData%\AutopilotMonitor\Config");
             _cacheFilePath = Path.Combine(cacheDir, "remote-config.json");
@@ -74,6 +76,14 @@ namespace AutopilotMonitor.Agent.Core.Configuration
             catch (Exception ex)
             {
                 _logger.Warning($"Failed to fetch remote config: {ex.Message}");
+
+                // Report to emergency channel if available (not available in --run-gather-rules mode)
+                if (_emergencyReporter != null)
+                {
+                    _ = _emergencyReporter.TrySendAsync(
+                        AgentErrorType.ConfigFetchFailed,
+                        ex.Message);
+                }
             }
 
             // Fall back to cached config

src/Agent/AutopilotMonitor.Agent.Core/Monitoring/Core/DiagnosticsPackageService.cs

@@ -304,8 +304,17 @@ namespace AutopilotMonitor.Agent.Core.Monitoring.Core
                         return (true, null);
                     }
 
-                    lastErrorCode = $"HTTP {(int)response.StatusCode} {response.ReasonPhrase}";
+                    var statusCode = (int)response.StatusCode;
+                    lastErrorCode = $"HTTP {statusCode} {response.ReasonPhrase}";
                     var responseBody = await response.Content.ReadAsStringAsync();
+
+                    // Auth errors (401/403) are permanent — SAS token invalid or expired, retrying won't help
+                    if (statusCode == 401 || statusCode == 403)
+                    {
+                        _logger.Warning($"Blob upload auth error (not retryable): {lastErrorCode} - {responseBody}");
+                        return (false, lastErrorCode);
+                    }
+
                     _logger.Warning($"Blob upload attempt {attempt} failed: {lastErrorCode} - {responseBody}");
                 }
                 catch (Exception ex)

src/Agent/AutopilotMonitor.Agent.Core/Monitoring/Core/MonitoringService.cs

@@ -252,7 +252,7 @@ namespace AutopilotMonitor.Agent.Core.Monitoring.Core
         {
             try
             {
-                _remoteConfigService = new RemoteConfigService(_apiClient, _configuration.TenantId, _logger);
+                _remoteConfigService = new RemoteConfigService(_apiClient, _configuration.TenantId, _logger, _emergencyReporter);
                 _remoteConfigService.FetchConfigAsync().Wait(TimeSpan.FromSeconds(15));
                 ApplyRuntimeSettingsFromRemoteConfig();
             }
@@ -546,6 +546,12 @@ namespace AutopilotMonitor.Agent.Core.Monitoring.Core
             catch (Exception ex)
             {
                 _logger.Error("Failed to register session", ex);
+
+                // Report to emergency channel — one-shot, no threshold needed.
+                // EmergencyReporter dedup ensures RegisterSessionFailed:null sent only once per session.
+                _ = _emergencyReporter.TrySendAsync(
+                    AgentErrorType.RegisterSessionFailed,
+                    ex.Message);
             }
         }
 

src/Agent/AutopilotMonitor.Agent.Core/Monitoring/Network/BackendApiClient.cs

@@ -115,6 +115,8 @@ namespace AutopilotMonitor.Agent.Core.Monitoring.Network
             var ndjson = CreateNdjson(request);
             var compressedContent = CompressWithGzip(ndjson);
 
+            _logger?.Debug($"IngestEventsAsync: POST {url} — {request.Events.Count} events, {compressedContent.Length} bytes compressed");
+
             var httpRequest = new HttpRequestMessage(HttpMethod.Post, url)
             {
                 Content = new ByteArrayContent(compressedContent)
@@ -165,6 +167,8 @@ namespace AutopilotMonitor.Agent.Core.Monitoring.Network
             var httpRequest = new HttpRequestMessage(HttpMethod.Get, url);
             AddSecurityHeaders(httpRequest);
 
+            _logger?.Debug($"GetAgentConfigAsync: GET {url}");
+
             var sw = Stopwatch.StartNew();
             var failed = false;
             long bytesDown = 0;

src/Agent/AutopilotMonitor.Agent.Core/Monitoring/Network/EmergencyReporter.cs

@@ -17,7 +17,7 @@ namespace AutopilotMonitor.Agent.Core.Monitoring.Network
     ///
     /// A failure in this reporter is silently swallowed — it must never cascade into the main loop.
     /// </summary>
-    internal class EmergencyReporter
+    public class EmergencyReporter
     {
         /// <summary>
         /// Number of consecutive upload failures required before the first emergency report is sent.

src/Agent/AutopilotMonitor.Agent/buildcounter.txt

@@ -1 +1 @@
-159
+166