Styjun

I am writing a simple training console app that triggers a fatal error when a user enters just numbers (on purpose). I have a main logger made up of 3 other loggers: 1 database, 1 file, 1 event. I change the file and database verbosity dynamically with separate LoggingLevelSwitches. The File switch is working as expected. The MSSqlServer correctly sets the switch, but I can't get anything to actually log unless the switch is set to Information or lower. Then all of the Warning, Error, Fatal entries that were not logged before are correctly entered into the database.

How do I get the sqlserver sink to correctly log entries when the switch is at a higher verbosity level? Everything works fine at lower verbosities, but I don't want to fill up the db with info unless needed. There is no issue with File or Event loggers

*I have a dispose object that closes and flushes the log

Thanks for any help!

creation of db logger

return new LoggerConfiguration()
 .MinimumLevel.ControlledBy(gDBLevel)
 .WriteTo.Async(a => a.MSSqlServer(connectionString: pInitObj.DBConnectionString
 , tableName: pInitObj.DBTableName
 , batchPostingLimit: 50
 , period: TimeSpan.FromMilliseconds(10)
 , formatProvider: null
 , autoCreateSqlTable: false
 , columnOptions: tmpOptions
 , schemaName: pInitObj.DBSchema))
 .CreateLogger();

creation of complete logger

return new LoggerConfiguration()
 .Enrich.WithProperty(Constants.DefaultDataTags.AppName.ToString(), pInitObj.LogName.ToString())
 .Enrich.With<EventID>()
 .Enrich.With<HttpUserName>()
 .Enrich.With<HttpUserIsAuthenticated>()
 .Enrich.FromLogContext()
 .Enrich.WithMvcActionName()
 .Enrich.WithMvcControllerName()
 .Enrich.WithHttpRequestClientHostIP()
 .Enrich.WithHttpRequestUrl()
 .Enrich.WithHttpSessionId()
 .Enrich.WithHttpRequestType()
 .Enrich.WithHttpSessionId()
 .Enrich.WithHttpRequestUserAgent()
 .Enrich.WithEnvironmentUserName()
 .WriteTo.Logger(getEventLogger(pInitObj)) 
 .WriteTo.Logger(getFileLogger(pInitObj))
 .WriteTo.Logger(getDBLogger(pInitObj))


 .CreateLogger();

It appears that the issue had to do with timing. Since all of the loggers were async, the program was closing too fast for the logs to get written on flush. I am pretty sure that the main program thread closes which collapses all of the spawned child threads on the logger before the logger has had time to complete the async processes.

Adding a half a second threading pause on the dispose after the flush call seems to have worked. Might have to bump it up to a second later on but whatevs.