Batch ingestion not correctly handling OperationCancelledException which can cause the ingestion to never used incoming context tasks and hang. (#4792)

Backport of #4780

Author: ramonsmits

src/Particular.LicensingComponent/AuditThroughput/AuditThroughputCollectorHostedService.cs

@@ -40,7 +40,7 @@ protected override async Task ExecuteAsync(CancellationToken cancellationToken)
                 }
             } while (await timer.WaitForNextTickAsync(cancellationToken));
         }
-        catch (OperationCanceledException)
+        catch (OperationCanceledException) when (cancellationToken.IsCancellationRequested)
         {
             logger.LogInformation($"Stopping {nameof(AuditThroughputCollectorHostedService)}");
         }

src/Particular.LicensingComponent/BrokerThroughput/BrokerThroughputCollectorHostedService.cs

@@ -53,7 +53,7 @@ static ReadOnlyDictionary<string, string> LoadBrokerSettingValues(IEnumerable<Ke
                 }
             } while (await timer.WaitForNextTickAsync(stoppingToken));
         }
-        catch (OperationCanceledException)
+        catch (OperationCanceledException) when (stoppingToken.IsCancellationRequested)
         {
             logger.LogInformation($"Stopping {nameof(BrokerThroughputCollectorHostedService)}");
         }

src/ServiceControl.AcceptanceTesting/ScenarioWithEndpointBehaviorExtensions.cs

@@ -139,9 +139,9 @@ public override async Task Stop(CancellationToken cancellationToken = default)
                 {
                     await checkTask;
                 }
-                catch (OperationCanceledException)
+                catch (OperationCanceledException) when (cancellationToken.IsCancellationRequested)
                 {
-                    //Swallow
+                    // Even though we are stopping, ONLY swallow when OCE from callee to not hide any ungraceful stop errors
                 }
                 finally
                 {

src/ServiceControl.Audit/Auditing/AuditIngestion.cs

@@ -17,7 +17,7 @@
     using ServiceControl.Infrastructure.Metrics;
     using Transports;
 
-    class AuditIngestion : IHostedService
+    class AuditIngestion : BackgroundService
     {
         static readonly long FrequencyInMilliseconds = Stopwatch.Frequency / 1000;
 
@@ -59,23 +59,15 @@ public AuditIngestion(
 
             errorHandlingPolicy = new AuditIngestionFaultPolicy(failedImportsStorage, settings.LoggingSettings, OnCriticalError);
 
-            watchdog = new Watchdog("audit message ingestion", EnsureStarted, EnsureStopped, ingestionState.ReportError, ingestionState.Clear, settings.TimeToRestartAuditIngestionAfterFailure, logger);
-
-            ingestionWorker = Task.Run(() => Loop(), CancellationToken.None);
-        }
-
-        public Task StartAsync(CancellationToken _) => watchdog.Start(() => applicationLifetime.StopApplication());
-
-        public async Task StopAsync(CancellationToken cancellationToken)
-        {
-            await watchdog.Stop();
-            channel.Writer.Complete();
-            await ingestionWorker;
-
-            if (transportInfrastructure != null)
-            {
-                await transportInfrastructure.Shutdown(cancellationToken);
-            }
+            watchdog = new Watchdog(
+                "audit message ingestion",
+                EnsureStarted,
+                EnsureStopped,
+                ingestionState.ReportError,
+                ingestionState.Clear,
+                settings.TimeToRestartAuditIngestionAfterFailure,
+                logger
+            );
         }
 
         Task OnCriticalError(string failure, Exception exception)
@@ -139,7 +131,7 @@ async Task EnsureStarted(CancellationToken cancellationToken = default)
                     }
                     catch (OperationCanceledException e) when (e.CancellationToken == cancellationToken)
                     {
-                        // ignored
+                        logger.Info("StopReceive cancelled");
                     }
                 }
 
@@ -176,7 +168,7 @@ async Task EnsureStopped(CancellationToken cancellationToken = default)
             }
             catch (OperationCanceledException e) when (e.CancellationToken == cancellationToken)
             {
-                // ignored
+                logger.Info("StopReceive cancelled");
             }
             finally
             {
@@ -202,51 +194,86 @@ async Task OnMessage(MessageContext messageContext, CancellationToken cancellati
             await taskCompletionSource.Task;
         }
 
-        async Task Loop()
+        public override async Task StartAsync(CancellationToken cancellationToken)
         {
-            var contexts = new List<MessageContext>(transportSettings.MaxConcurrency.Value);
+            await watchdog.Start(() => applicationLifetime.StopApplication());
+            await base.StartAsync(cancellationToken);
+        }
 
-            while (await channel.Reader.WaitToReadAsync())
+        protected override async Task ExecuteAsync(CancellationToken stoppingToken)
+        {
+            try
             {
-                // will only enter here if there is something to read.
-                try
+                var contexts = new List<MessageContext>(transportSettings.MaxConcurrency.Value);
+
+                while (await channel.Reader.WaitToReadAsync(stoppingToken))
                 {
-                    // as long as there is something to read this will fetch up to MaximumConcurrency items
-                    while (channel.Reader.TryRead(out var context))
+                    // will only enter here if there is something to read.
+                    try
                     {
-                        contexts.Add(context);
+                        // as long as there is something to read this will fetch up to MaximumConcurrency items
+                        while (channel.Reader.TryRead(out var context))
+                        {
+                            contexts.Add(context);
+                        }
+
+                        batchSizeMeter.Mark(contexts.Count);
+                        using (batchDurationMeter.Measure())
+                        {
+                            await auditIngestor.Ingest(contexts);
+                        }
                     }
+                    catch (Exception e)
+                    {
+                        // signal all message handling tasks to terminate
+                        foreach (var context in contexts)
+                        {
+                            _ = context.GetTaskCompletionSource().TrySetException(e);
+                        }
+
+                        if (e is OperationCanceledException && stoppingToken.IsCancellationRequested)
+                        {
+                            logger.Info("Batch cancelled", e);
+                            break;
+                        }
 
-                    batchSizeMeter.Mark(contexts.Count);
-                    using (batchDurationMeter.Measure())
+                        logger.Info("Ingesting messages failed", e);
+                    }
+                    finally
                     {
-                        await auditIngestor.Ingest(contexts);
+                        contexts.Clear();
                     }
                 }
-                catch (OperationCanceledException)
-                {
-                    //Do nothing as we are shutting down
-                    continue;
-                }
-                catch (Exception e) // show must go on
+                // will fall out here when writer is completed
+            }
+            catch (OperationCanceledException) when (stoppingToken.IsCancellationRequested)
+            {
+                // ExecuteAsync cancelled
+            }
+        }
+
+        public override async Task StopAsync(CancellationToken cancellationToken)
+        {
+            try
+            {
+                await watchdog.Stop();
+                channel.Writer.Complete();
+                await base.StopAsync(cancellationToken);
+            }
+            finally
+            {
+                if (transportInfrastructure != null)
                 {
-                    if (logger.IsInfoEnabled)
+                    try
                     {
-                        logger.Info("Ingesting messages failed", e);
+                        await transportInfrastructure.Shutdown(cancellationToken);
                     }
-
-                    // signal all message handling tasks to terminate
-                    foreach (var context in contexts)
+                    catch (OperationCanceledException e) when (cancellationToken.IsCancellationRequested)
                     {
-                        context.GetTaskCompletionSource().TrySetException(e);
+                        logger.Info("Shutdown cancelled", e);
                     }
                 }
-                finally
-                {
-                    contexts.Clear();
-                }
             }
-            // will fall out here when writer is completed
         }
 
         TransportInfrastructure transportInfrastructure;
@@ -265,7 +292,6 @@ async Task Loop()
         readonly Meter batchDurationMeter;
         readonly Counter receivedMeter;
         readonly Watchdog watchdog;
-        readonly Task ingestionWorker;
         readonly IHostApplicationLifetime applicationLifetime;
 
         static readonly ILog logger = LogManager.GetLogger<AuditIngestion>();

src/ServiceControl.Audit/Auditing/ImportFailedAudits.cs

@@ -48,9 +48,9 @@ await failedAuditStore.ProcessFailedMessages(
                                 Logger.Debug($"Successfully re-imported failed audit message {transportMessage.Id}.");
                             }
                         }
-                        catch (OperationCanceledException)
+                        catch (OperationCanceledException e) when (token.IsCancellationRequested)
                         {
-                            // no-op
+                            Logger.Info("Cancelled", e);
                         }
                         catch (Exception e)
                         {

src/ServiceControl.Audit/Infrastructure/Hosting/Commands/ImportFailedAuditsCommand.cs

@@ -7,10 +7,13 @@
     using Microsoft.Extensions.DependencyInjection;
     using Microsoft.Extensions.Hosting;
     using NServiceBus;
+    using NServiceBus.Logging;
     using Settings;
 
     class ImportFailedAuditsCommand : AbstractCommand
     {
+        readonly ILog logger = LogManager.GetLogger<ImportFailedAuditsCommand>();
+
         public override async Task Execute(HostArguments args, Settings settings)
         {
             settings.IngestAuditMessages = false;
@@ -37,9 +40,9 @@ public override async Task Execute(HostArguments args, Settings settings)
             {
                 await importer.Run(tokenSource.Token);
             }
-            catch (OperationCanceledException)
+            catch (OperationCanceledException e) when (tokenSource.IsCancellationRequested)
             {
-                // no op
+                logger.Info("Cancelled", e);
             }
             finally
             {

src/ServiceControl.Infrastructure.Metrics/MetricsReporter.cs

@@ -32,7 +32,7 @@ public void Start()
                         await Task.Delay(interval, tokenSource.Token).ConfigureAwait(false);
                     }
                 }
-                catch (OperationCanceledException)
+                catch (OperationCanceledException) when (tokenSource.IsCancellationRequested)
                 {
                     //no-op
                 }

src/ServiceControl.Infrastructure/AsyncTimer.cs

@@ -40,17 +40,17 @@ public TimerJob(Func<CancellationToken, Task<TimerJobExecutionResult>> callback,
 
                             //Otherwise execute immediately
                         }
-                        catch (OperationCanceledException)
+                        catch (OperationCanceledException) when (token.IsCancellationRequested)
                         {
-                            // no-op
+                            break;
                         }
                         catch (Exception ex)
                         {
                             errorCallback(ex);
                         }
                     }
                 }
-                catch (OperationCanceledException)
+                catch (OperationCanceledException) when (token.IsCancellationRequested)
                 {
                     // no-op
                 }
@@ -64,7 +64,7 @@ public async Task Stop()
                 return;
             }
 
-            tokenSource.Cancel();
+            await tokenSource.CancelAsync().ConfigureAwait(false);
             tokenSource.Dispose();
 
             if (task != null)
@@ -73,7 +73,7 @@ public async Task Stop()
                 {
                     await task.ConfigureAwait(false);
                 }
-                catch (OperationCanceledException)
+                catch (OperationCanceledException) when (tokenSource.IsCancellationRequested)
                 {
                     //NOOP
                 }

src/ServiceControl.Infrastructure/ReadOnlyStream.cs

@@ -51,7 +51,7 @@ public override Task CopyToAsync(Stream destination, int bufferSize, Cancellatio
 
             return Task.CompletedTask;
         }
-        catch (OperationCanceledException e)
+        catch (OperationCanceledException e) when (cancellationToken.IsCancellationRequested)
         {
             return Task.FromCanceled(e.CancellationToken);
         }
@@ -113,7 +113,7 @@ public override Task<int> ReadAsync(byte[] buffer, int offset, int count, Cancel
 
             return Task.FromResult(result);
         }
-        catch (OperationCanceledException e)
+        catch (OperationCanceledException e) when (cancellationToken.IsCancellationRequested)
         {
             return Task.FromCanceled<int>(e.CancellationToken);
         }
@@ -136,7 +136,7 @@ public override ValueTask<int> ReadAsync(Memory<byte> buffer, CancellationToken
 
             return new ValueTask<int>(result);
         }
-        catch (OperationCanceledException e)
+        catch (OperationCanceledException e) when (cancellationToken.IsCancellationRequested)
         {
             return new ValueTask<int>(Task.FromCanceled<int>(e.CancellationToken));
         }

src/ServiceControl.Infrastructure/Watchdog.cs

@@ -54,9 +54,10 @@ public Task Start(Action onFailedOnStartup)
 
                         failedOnStartup ??= false;
                     }
-                    catch (OperationCanceledException)
+                    catch (OperationCanceledException e) when (!shutdownTokenSource.IsCancellationRequested)
                     {
-                        //Do not Delay
+                        // Continue, as OCE is not from caller
+                        log.Info("Start cancelled, retrying...", e);
                         continue;
                     }
                     catch (Exception e)
@@ -81,9 +82,9 @@ public Task Start(Action onFailedOnStartup)
                     {
                         await Task.Delay(timeToWaitBetweenStartupAttempts, shutdownTokenSource.Token).ConfigureAwait(false);
                     }
-                    catch (OperationCanceledException)
+                    catch (OperationCanceledException) when (shutdownTokenSource.IsCancellationRequested)
                     {
-                        //Ignore
+                        //Ignore, no need to log cancellation of delay
                     }
                 }
                 try