Make CreateAzureCdnWarehouseReports SQL timeout configurable (#283)

Scott Bommarito · web-flow · commit c6bd2f6cce6d · 2017-12-13T11:18:56.000-08:00
diff --git a/src/Stats.CreateAzureCdnWarehouseReports/Job.cs b/src/Stats.CreateAzureCdnWarehouseReports/Job.cs
@@ -17,6 +17,7 @@ namespace Stats.CreateAzureCdnWarehouseReports
     public class Job
         : JobBase
     {
+        private const int DefaultSqlCommandTimeout = 1800; // 30 minute SQL command timeout by default
         private const string _recentPopularityDetailByPackageReportBaseName = "recentpopularitydetail_";
         private CloudStorageAccount _cloudStorageAccount;
         private CloudStorageAccount _dataStorageAccount;
@@ -25,6 +26,7 @@ public class Job
         private SqlConnectionStringBuilder _galleryDatabase;
         private string _reportName;
         private string[] _dataContainerNames;
+        private int _sqlCommandTimeout = DefaultSqlCommandTimeout;
 
         private static readonly IDictionary<string, string> _storedProcedures = new Dictionary<string, string>
         {
@@ -48,6 +50,7 @@ public override void Init(IDictionary<string, string> jobArgsDictionary)
             var statisticsDatabaseConnectionString = JobConfigurationManager.GetArgument(jobArgsDictionary, JobArgumentNames.StatisticsDatabase);
             var galleryDatabaseConnectionString = JobConfigurationManager.GetArgument(jobArgsDictionary, JobArgumentNames.SourceDatabase);
             var dataStorageAccountConnectionString = JobConfigurationManager.GetArgument(jobArgsDictionary, JobArgumentNames.DataStorageAccount);
+            _sqlCommandTimeout = JobConfigurationManager.TryGetIntArgument(jobArgsDictionary, JobArgumentNames.CommandTimeOut) ?? DefaultSqlCommandTimeout;
 
             _cloudStorageAccount = ValidateAzureCloudStorageAccount(cloudStorageAccountConnectionString, JobArgumentNames.AzureCdnCloudStorageAccount);
             _statisticsContainerName = ValidateAzureContainerName(JobConfigurationManager.GetArgument(jobArgsDictionary, JobArgumentNames.AzureCdnCloudStorageContainerName), JobArgumentNames.AzureCdnCloudStorageContainerName);
@@ -81,12 +84,12 @@ public override async Task Run()
                 // generate all reports
                 var reportGenerators = new Dictionary<ReportBuilder, ReportDataCollector>
                     {
-                        { new ReportBuilder(reportBuilderLogger, ReportNames.NuGetClientVersion), new ReportDataCollector(reportCollectorLogger, _storedProcedures[ReportNames.NuGetClientVersion], _statisticsDatabase) },
-                        { new ReportBuilder(reportBuilderLogger, ReportNames.Last6Weeks), new ReportDataCollector(reportCollectorLogger, _storedProcedures[ReportNames.Last6Weeks], _statisticsDatabase) },
-                        { new ReportBuilder(reportBuilderLogger, ReportNames.RecentCommunityPopularity), new ReportDataCollector(reportCollectorLogger, _storedProcedures[ReportNames.RecentCommunityPopularity], _statisticsDatabase) },
-                        { new ReportBuilder(reportBuilderLogger, ReportNames.RecentCommunityPopularityDetail), new ReportDataCollector(reportCollectorLogger, _storedProcedures[ReportNames.RecentCommunityPopularityDetail], _statisticsDatabase) },
-                        { new ReportBuilder(reportBuilderLogger, ReportNames.RecentPopularity), new ReportDataCollector(reportCollectorLogger, _storedProcedures[ReportNames.RecentPopularity], _statisticsDatabase) },
-                        { new ReportBuilder(reportBuilderLogger, ReportNames.RecentPopularityDetail), new ReportDataCollector(reportCollectorLogger, _storedProcedures[ReportNames.RecentPopularityDetail], _statisticsDatabase) }
+                        { new ReportBuilder(reportBuilderLogger, ReportNames.NuGetClientVersion), new ReportDataCollector(reportCollectorLogger, _storedProcedures[ReportNames.NuGetClientVersion], _statisticsDatabase, _sqlCommandTimeout) },
+                        { new ReportBuilder(reportBuilderLogger, ReportNames.Last6Weeks), new ReportDataCollector(reportCollectorLogger, _storedProcedures[ReportNames.Last6Weeks], _statisticsDatabase, _sqlCommandTimeout) },
+                        { new ReportBuilder(reportBuilderLogger, ReportNames.RecentCommunityPopularity), new ReportDataCollector(reportCollectorLogger, _storedProcedures[ReportNames.RecentCommunityPopularity], _statisticsDatabase, _sqlCommandTimeout) },
+                        { new ReportBuilder(reportBuilderLogger, ReportNames.RecentCommunityPopularityDetail), new ReportDataCollector(reportCollectorLogger, _storedProcedures[ReportNames.RecentCommunityPopularityDetail], _statisticsDatabase, _sqlCommandTimeout) },
+                        { new ReportBuilder(reportBuilderLogger, ReportNames.RecentPopularity), new ReportDataCollector(reportCollectorLogger, _storedProcedures[ReportNames.RecentPopularity], _statisticsDatabase, _sqlCommandTimeout) },
+                        { new ReportBuilder(reportBuilderLogger, ReportNames.RecentPopularityDetail), new ReportDataCollector(reportCollectorLogger, _storedProcedures[ReportNames.RecentPopularityDetail], _statisticsDatabase, _sqlCommandTimeout) }
                     };
 
                 foreach (var reportGenerator in reportGenerators)
@@ -102,7 +105,7 @@ public override async Task Run()
             {
                 // generate only the specific report
                 var reportBuilder = new ReportBuilder(reportBuilderLogger, _reportName);
-                var reportDataCollector = new ReportDataCollector(reportCollectorLogger, _storedProcedures[_reportName], _statisticsDatabase);
+                var reportDataCollector = new ReportDataCollector(reportCollectorLogger, _storedProcedures[_reportName], _statisticsDatabase, _sqlCommandTimeout);
 
                 await ProcessReport(LoggerFactory, destinationContainer, reportBuilder, reportDataCollector, reportGenerationTime);
             }
@@ -162,14 +165,14 @@ private static async Task ProcessReport(ILoggerFactory loggerFactory, CloudBlobC
 
         private async Task RebuildPackageReports(CloudBlobContainer destinationContainer, DateTime reportGenerationTime)
         {
-            var dirtyPackageIds = await ReportDataCollector.GetDirtyPackageIds(LoggerFactory.CreateLogger<ReportDataCollector>(), _statisticsDatabase, reportGenerationTime);
+            var dirtyPackageIds = await ReportDataCollector.GetDirtyPackageIds(LoggerFactory.CreateLogger<ReportDataCollector>(), _statisticsDatabase, reportGenerationTime, _sqlCommandTimeout);
 
             if (!dirtyPackageIds.Any())
                 return;
 
             // first process the top 100 packages
             var top100 = dirtyPackageIds.Take(100);
-            var reportDataCollector = new ReportDataCollector(LoggerFactory.CreateLogger<ReportDataCollector>(), _storedProceduresPerPackageId[ReportNames.RecentPopularityDetailByPackageId], _statisticsDatabase);
+            var reportDataCollector = new ReportDataCollector(LoggerFactory.CreateLogger<ReportDataCollector>(), _storedProceduresPerPackageId[ReportNames.RecentPopularityDetailByPackageId], _statisticsDatabase, _sqlCommandTimeout);
             var top100Task = Parallel.ForEach(top100, new ParallelOptions { MaxDegreeOfParallelism = 4 }, dirtyPackageId =>
             {
                 var packageId = dirtyPackageId.PackageId.ToLowerInvariant();
@@ -199,7 +202,8 @@ private async Task RebuildPackageReports(CloudBlobContainer destinationContainer
                                 new ReportDataCollector(
                                     LoggerFactory.CreateLogger<ReportDataCollector>(),
                                     _storedProceduresPerPackageId[ReportNames.RecentPopularityDetailByPackageId],
-                                    _statisticsDatabase)
+                                    _statisticsDatabase,
+                                    _sqlCommandTimeout)
                             }
                         };
 
@@ -215,15 +219,15 @@ private async Task RebuildPackageReports(CloudBlobContainer destinationContainer
                 if (top100Task.IsCompleted)
                 {
                     var runToCursor = dirtyPackageIds.First().RunToCuror;
-                    await ReportDataCollector.UpdateDirtyPackageIdCursor(_statisticsDatabase, runToCursor);
+                    await ReportDataCollector.UpdateDirtyPackageIdCursor(_statisticsDatabase, runToCursor, _sqlCommandTimeout);
                 }
             }
         }
 
         private async Task CleanInactiveRecentPopularityDetailByPackageReports(CloudBlobContainer destinationContainer, DateTime reportGenerationTime)
         {
             Logger.LogDebug("Getting list of inactive packages.");
-            var packageIds = await ReportDataCollector.ListInactivePackageIdReports(_statisticsDatabase, reportGenerationTime);
+            var packageIds = await ReportDataCollector.ListInactivePackageIdReports(_statisticsDatabase, reportGenerationTime, _sqlCommandTimeout);
             Logger.LogInformation("Found {InactivePackageCount} inactive packages.", packageIds.Count);
 
             // Collect the list of reports
diff --git a/src/Stats.CreateAzureCdnWarehouseReports/ReportDataCollector.cs b/src/Stats.CreateAzureCdnWarehouseReports/ReportDataCollector.cs
@@ -13,17 +13,18 @@ namespace Stats.CreateAzureCdnWarehouseReports
 {
     internal class ReportDataCollector
     {
-        private const int _commandTimeout = 1800; // 30 minutes max
+        private int _commandTimeout;
         private readonly string _procedureName;
         private readonly SqlConnectionStringBuilder _sourceDatabase;
 
         private ILogger<ReportDataCollector> _logger;
 
-        public ReportDataCollector(ILogger<ReportDataCollector> logger, string procedureName, SqlConnectionStringBuilder sourceDatabase)
+        public ReportDataCollector(ILogger<ReportDataCollector> logger, string procedureName, SqlConnectionStringBuilder sourceDatabase, int timeout)
         {
             _logger = logger;
             _procedureName = procedureName;
             _sourceDatabase = sourceDatabase;
+            _commandTimeout = timeout;
         }
 
         public async Task<DataTable> CollectAsync(DateTime reportGenerationTime, params Tuple<string, int, string>[] parameters)
@@ -40,27 +41,27 @@ public async Task<DataTable> CollectAsync(DateTime reportGenerationTime, params
             return table;
         }
 
-        public static async Task<IReadOnlyCollection<DirtyPackageId>> GetDirtyPackageIds(ILogger logger, SqlConnectionStringBuilder sourceDatabase, DateTime reportGenerationTime)
+        public static async Task<IReadOnlyCollection<DirtyPackageId>> GetDirtyPackageIds(ILogger logger, SqlConnectionStringBuilder sourceDatabase, DateTime reportGenerationTime, int commandTimeout)
         {
             logger.LogInformation("Getting list of dirty packages IDs.");
 
             IReadOnlyCollection<DirtyPackageId> packageIds = new List<DirtyPackageId>();
 
             // Get the data
-            await WithRetry(async () => packageIds = await GetDirtyPackageIdsFromWarehouse(sourceDatabase, reportGenerationTime), logger);
+            await WithRetry(async () => packageIds = await GetDirtyPackageIdsFromWarehouse(sourceDatabase, reportGenerationTime, commandTimeout), logger);
 
             logger.LogInformation("Found {DirtyPackagesCount} dirty packages to update.", packageIds.Count);
 
             return packageIds;
         }
 
-        public static async Task<IReadOnlyCollection<string>> ListInactivePackageIdReports(SqlConnectionStringBuilder sourceDatabase, DateTime reportGenerationTime)
+        public static async Task<IReadOnlyCollection<string>> ListInactivePackageIdReports(SqlConnectionStringBuilder sourceDatabase, DateTime reportGenerationTime, int commandTimeout)
         {
             using (var connection = await sourceDatabase.ConnectTo())
             {
                 var command = new SqlCommand("[dbo].[DownloadReportListInactive]", connection);
                 command.CommandType = CommandType.StoredProcedure;
-                command.CommandTimeout = _commandTimeout;
+                command.CommandTimeout = commandTimeout;
 
                 command.Parameters.Add("ReportGenerationTime", SqlDbType.DateTime).Value = reportGenerationTime;
 
@@ -132,13 +133,13 @@ private async Task<DataTable> ExecuteSql(DateTime reportGenerationTime, params T
             }
         }
 
-        private static async Task<IReadOnlyCollection<DirtyPackageId>> GetDirtyPackageIdsFromWarehouse(SqlConnectionStringBuilder sourceDatabase, DateTime reportGenerationTime)
+        private static async Task<IReadOnlyCollection<DirtyPackageId>> GetDirtyPackageIdsFromWarehouse(SqlConnectionStringBuilder sourceDatabase, DateTime reportGenerationTime, int commandTimeout)
         {
             using (var connection = await sourceDatabase.ConnectTo())
             {
                 var command = new SqlCommand("[dbo].[GetDirtyPackageIds]", connection);
                 command.CommandType = CommandType.StoredProcedure;
-                command.CommandTimeout = _commandTimeout;
+                command.CommandTimeout = commandTimeout;
 
                 command.Parameters.Add("ReportGenerationTime", SqlDbType.DateTime).Value = reportGenerationTime;
 
@@ -155,13 +156,13 @@ private static async Task<IReadOnlyCollection<DirtyPackageId>> GetDirtyPackageId
             }
         }
 
-        public static async Task UpdateDirtyPackageIdCursor(SqlConnectionStringBuilder sourceDatabase, DateTime runToCursor)
+        public static async Task UpdateDirtyPackageIdCursor(SqlConnectionStringBuilder sourceDatabase, DateTime runToCursor, int commandTimeout)
         {
             using (var connection = await sourceDatabase.ConnectTo())
             {
                 var command = new SqlCommand("[dbo].[UpdateDirtyPackageIdCursor]", connection);
                 command.CommandType = CommandType.StoredProcedure;
-                command.CommandTimeout = _commandTimeout;
+                command.CommandTimeout = commandTimeout;
                 command.Parameters.Add("@Position", SqlDbType.DateTime).Value = runToCursor;
 
                 await command.ExecuteNonQueryAsync();
diff --git a/src/Stats.CreateAzureCdnWarehouseReports/Scripts/Stats.CreateAzureCdnWarehouseReports.cmd b/src/Stats.CreateAzureCdnWarehouseReports/Scripts/Stats.CreateAzureCdnWarehouseReports.cmd
@@ -7,7 +7,7 @@ cd bin
 
 	title #{Jobs.stats.createazurecdnwarehousereports.Title}
 
-	start /w stats.createazurecdnwarehousereports.exe -VaultName "#{Deployment.Azure.KeyVault.VaultName}" -ClientId "#{Deployment.Azure.KeyVault.ClientId}" -CertificateThumbprint "#{Deployment.Azure.KeyVault.CertificateThumbprint}" -AzureCdnCloudStorageAccount "#{Jobs.stats.createazurecdnwarehousereports.AzureCdn.CloudStorageAccount}" -AzureCdnCloudStorageContainerName "#{Jobs.stats.createazurecdnwarehousereports.AzureCdn.CloudStorageContainerName}" -StatisticsDatabase "#{Jobs.stats.createazurecdnwarehousereports.StatisticsDatabase}" -SourceDatabase "#{Jobs.stats.createazurecdnwarehousereports.SourceDatabase}" -DataStorageAccount "#{Jobs.stats.createazurecdnwarehousereports.DataStorageAccount}" -InstrumentationKey "#{Jobs.stats.createazurecdnwarehousereports.InstrumentationKey}" -DataContainerName "#{Jobs.stats.createazurecdnwarehousereports.DataContainerName}" -verbose true -Interval #{Jobs.stats.createazurecdnwarehousereports.Interval} 
+	start /w stats.createazurecdnwarehousereports.exe -VaultName "#{Deployment.Azure.KeyVault.VaultName}" -ClientId "#{Deployment.Azure.KeyVault.ClientId}" -CertificateThumbprint "#{Deployment.Azure.KeyVault.CertificateThumbprint}" -AzureCdnCloudStorageAccount "#{Jobs.stats.createazurecdnwarehousereports.AzureCdn.CloudStorageAccount}" -AzureCdnCloudStorageContainerName "#{Jobs.stats.createazurecdnwarehousereports.AzureCdn.CloudStorageContainerName}" -StatisticsDatabase "#{Jobs.stats.createazurecdnwarehousereports.StatisticsDatabase}" -SourceDatabase "#{Jobs.stats.createazurecdnwarehousereports.SourceDatabase}" -DataStorageAccount "#{Jobs.stats.createazurecdnwarehousereports.DataStorageAccount}" -InstrumentationKey "#{Jobs.stats.createazurecdnwarehousereports.InstrumentationKey}" -DataContainerName "#{Jobs.stats.createazurecdnwarehousereports.DataContainerName}" -CommandTimeOut "#{Jobs.stats.createazurecdnwarehousereports.CommandTimeOut}" -verbose true -Interval #{Jobs.stats.createazurecdnwarehousereports.Interval} 
 
 	echo "Finished #{Jobs.stats.createazurecdnwarehousereports.Title}"
 

Original file line number	Diff line number	Diff line change
`@@ -13,17 +13,18 @@ namespace Stats.CreateAzureCdnWarehouseReports`
`13`	`13`	`{`
`14`	`14`	`internal class ReportDataCollector`
`15`	`15`	`{`
`16`		`- private const int _commandTimeout = 1800; // 30 minutes max`
	`16`	`+ private int _commandTimeout;`
`17`	`17`	`private readonly string _procedureName;`
`18`	`18`	`private readonly SqlConnectionStringBuilder _sourceDatabase;`
`19`	`19`
`20`	`20`	`private ILogger<ReportDataCollector> _logger;`
`21`	`21`
`22`		`- public ReportDataCollector(ILogger<ReportDataCollector> logger, string procedureName, SqlConnectionStringBuilder sourceDatabase)`
	`22`	`+ public ReportDataCollector(ILogger<ReportDataCollector> logger, string procedureName, SqlConnectionStringBuilder sourceDatabase, int timeout)`
`23`	`23`	`{`
`24`	`24`	`_logger = logger;`
`25`	`25`	`_procedureName = procedureName;`
`26`	`26`	`_sourceDatabase = sourceDatabase;`
	`27`	`+ _commandTimeout = timeout;`
`27`	`28`	`}`
`28`	`29`
`29`	`30`	`public async Task<DataTable> CollectAsync(DateTime reportGenerationTime, params Tuple<string, int, string>[] parameters)`
`@@ -40,27 +41,27 @@ public async Task<DataTable> CollectAsync(DateTime reportGenerationTime, params`
`40`	`41`	`return table;`
`41`	`42`	`}`
`42`	`43`
`43`		`- public static async Task<IReadOnlyCollection<DirtyPackageId>> GetDirtyPackageIds(ILogger logger, SqlConnectionStringBuilder sourceDatabase, DateTime reportGenerationTime)`
	`44`	`+ public static async Task<IReadOnlyCollection<DirtyPackageId>> GetDirtyPackageIds(ILogger logger, SqlConnectionStringBuilder sourceDatabase, DateTime reportGenerationTime, int commandTimeout)`
`44`	`45`	`{`
`45`	`46`	`logger.LogInformation("Getting list of dirty packages IDs.");`
`46`	`47`
`47`	`48`	`IReadOnlyCollection<DirtyPackageId> packageIds = new List<DirtyPackageId>();`
`48`	`49`
`49`	`50`	`// Get the data`
`50`		`- await WithRetry(async () => packageIds = await GetDirtyPackageIdsFromWarehouse(sourceDatabase, reportGenerationTime), logger);`
	`51`	`+ await WithRetry(async () => packageIds = await GetDirtyPackageIdsFromWarehouse(sourceDatabase, reportGenerationTime, commandTimeout), logger);`
`51`	`52`
`52`	`53`	`logger.LogInformation("Found {DirtyPackagesCount} dirty packages to update.", packageIds.Count);`
`53`	`54`
`54`	`55`	`return packageIds;`
`55`	`56`	`}`
`56`	`57`
`57`		`- public static async Task<IReadOnlyCollection<string>> ListInactivePackageIdReports(SqlConnectionStringBuilder sourceDatabase, DateTime reportGenerationTime)`
	`58`	`+ public static async Task<IReadOnlyCollection<string>> ListInactivePackageIdReports(SqlConnectionStringBuilder sourceDatabase, DateTime reportGenerationTime, int commandTimeout)`
`58`	`59`	`{`
`59`	`60`	`using (var connection = await sourceDatabase.ConnectTo())`
`60`	`61`	`{`
`61`	`62`	`var command = new SqlCommand("[dbo].[DownloadReportListInactive]", connection);`
`62`	`63`	`command.CommandType = CommandType.StoredProcedure;`
`63`		`- command.CommandTimeout = _commandTimeout;`
	`64`	`+ command.CommandTimeout = commandTimeout;`
`64`	`65`
`65`	`66`	`command.Parameters.Add("ReportGenerationTime", SqlDbType.DateTime).Value = reportGenerationTime;`
`66`	`67`
`@@ -132,13 +133,13 @@ private async Task<DataTable> ExecuteSql(DateTime reportGenerationTime, params T`
`132`	`133`	`}`
`133`	`134`	`}`
`134`	`135`
`135`		`- private static async Task<IReadOnlyCollection<DirtyPackageId>> GetDirtyPackageIdsFromWarehouse(SqlConnectionStringBuilder sourceDatabase, DateTime reportGenerationTime)`
	`136`	`+ private static async Task<IReadOnlyCollection<DirtyPackageId>> GetDirtyPackageIdsFromWarehouse(SqlConnectionStringBuilder sourceDatabase, DateTime reportGenerationTime, int commandTimeout)`
`136`	`137`	`{`
`137`	`138`	`using (var connection = await sourceDatabase.ConnectTo())`
`138`	`139`	`{`
`139`	`140`	`var command = new SqlCommand("[dbo].[GetDirtyPackageIds]", connection);`
`140`	`141`	`command.CommandType = CommandType.StoredProcedure;`
`141`		`- command.CommandTimeout = _commandTimeout;`
	`142`	`+ command.CommandTimeout = commandTimeout;`
`142`	`143`
`143`	`144`	`command.Parameters.Add("ReportGenerationTime", SqlDbType.DateTime).Value = reportGenerationTime;`
`144`	`145`
`@@ -155,13 +156,13 @@ private static async Task<IReadOnlyCollection<DirtyPackageId>> GetDirtyPackageId`
`155`	`156`	`}`
`156`	`157`	`}`
`157`	`158`
`158`		`- public static async Task UpdateDirtyPackageIdCursor(SqlConnectionStringBuilder sourceDatabase, DateTime runToCursor)`
	`159`	`+ public static async Task UpdateDirtyPackageIdCursor(SqlConnectionStringBuilder sourceDatabase, DateTime runToCursor, int commandTimeout)`
`159`	`160`	`{`
`160`	`161`	`using (var connection = await sourceDatabase.ConnectTo())`
`161`	`162`	`{`
`162`	`163`	`var command = new SqlCommand("[dbo].[UpdateDirtyPackageIdCursor]", connection);`
`163`	`164`	`command.CommandType = CommandType.StoredProcedure;`
`164`		`- command.CommandTimeout = _commandTimeout;`
	`165`	`+ command.CommandTimeout = commandTimeout;`
`165`	`166`	`command.Parameters.Add("@Position", SqlDbType.DateTime).Value = runToCursor;`
`166`	`167`
`167`	`168`	`await command.ExecuteNonQueryAsync();`