Saturday, 28 June 2025

Setting up connection resiliency for Entity Framework

In Entity Framework, it is possible to add more connection resiliency. This can be done for example if you are working against a more unstable database connection, maybe because the database is served in the Cloud and/or is not scaled properly to its load. Whatever reason, it is possible to add more connection resiliency. The connection resiliency can be used in other scenarios that just SQL servers hosted in Azure, such as On-Premise databases. It should add more resiliency and stability for scenarios where connections to database needs to be improved. This could also be due to mobile clients being moved in and out of areas with good network access, such as within buildings and factories on different levels trying to access a wireless connection that connects to a database. The code in this article is available in my Github repo here:

https://github.com/toreaurstadboss/BulkOperationsEntityFramework

first off, the ExecutionStrategy is set up. A DbConfiguration is added to set this up.

ApplicationDbModelConfiguration.cs



using System;
using System.Data.Entity;
using System.Data.Entity.SqlServer;

namespace BulkOperationsEntityFramework
{
    public class ApplicationDbConfiguration : DbConfiguration
    {

        public ApplicationDbConfiguration()
        {
            SetExecutionStrategy(SqlProviderServices.ProviderInvariantName, () =>
             new CustomSqlAzureExecutionStrategy(maxRetryCount: 10, maxDelay: TimeSpan.FromSeconds(5))); //note : max total delay of retries is 30 seconds per default in SQL Server
        }

    }

}


In EF Core 8 (using Entity Framework with .NET 8, you can set it up like this :

Program.cs




    public class ApplicationDbContext : DbContext
    {
        protected override void OnConfiguring(DbContextOptionsBuilder optionsBuilder)
        {
            optionsBuilder.UseSqlServer(
                "DefaultConnection",
                sqlOptions =>
                {
                    sqlOptions.EnableRetryOnFailure(
                        maxRetryCount: 10,
                        maxRetryDelay: TimeSpan.FromSeconds(5),
                        errorNumbersToAdd: null
                    );
                });
        }



Setting up the interval strategy

The CustomSqlAzureExecutionStrategy inherits from the SqlAzureExecutionStrategy. delay = min ( maxDelay , random × ( 2 retryCount 1 ) × baseDelay ) The default base delay in Entity Framework is to wait one second, so the next wait time will be about 2 seconds, then the next delays will quickly grow up the max wait time of five seconds. The custom sql azure execution strategy implementation inherits from SqlAzureExecutionStrategy.

CustomSqlAzureExecutionStrategy.cs



using System;
using System.Data.Entity.SqlServer;

namespace BulkOperationsEntityFramework
{

    public class CustomSqlAzureExecutionStrategy : SqlAzureExecutionStrategy
    {

        [ThreadStatic]
        private static int _currentRetryCount = 0;

        public CustomSqlAzureExecutionStrategy(int maxRetryCount, TimeSpan maxDelay)
        : base(maxRetryCount, maxDelay) { }

        protected override bool ShouldRetryOn(Exception ex)
        {
            _currentRetryCount++;
            Console.WriteLine($"{nameof(CustomSqlAzureExecutionStrategy)}: Retry-count within thread: {_currentRetryCount}");
            Log.Information("{Class}: Retry-count within thread: {RetryCount} {ExceptionType}", nameof(CustomSqlAzureExecutionStrategy), _currentRetryCount, ex.GetType().Name);

            return base.ShouldRetryOn(ex) || ex is SimulatedTransientSqlException;
        }

    }

}


Of course, just logging out to console probably is not a very elegant solution, and it could instead be logged out to for example SeriLog, which is used in the line with the Log.Information call. The SimulatedTransientSqlException looks like this:

SimulatedTransientSqlException.cs



 public class SimulatedTransientSqlException : Exception
 {
     public SimulatedTransientSqlException()
     : base("Simulated transient SQL exception.") { }
 }
 

The following db interceptor is added to simulate transient failures happening, at 10% chance of it happening.

TransientFailureInterceptor.cs



using System;
using System.Data.Common;
using System.Data.Entity.Infrastructure.Interception;
using System.Diagnostics;

namespace BulkOperationsEntityFramework
{

    public class TransientFailureInterceptor : DbCommandInterceptor
    {
        private static readonly Random _random = new Random();

        public override void ReaderExecuting(DbCommand command, DbCommandInterceptionContext<DbDataReader> interceptionContext)
        {
            SimulateTransientFailure(interceptionContext);
            base.ReaderExecuting(command, interceptionContext);
        }

        public override void ScalarExecuting(DbCommand command, DbCommandInterceptionContext<object> interceptionContext)
        {
            SimulateTransientFailure(interceptionContext);
            base.ScalarExecuting(command, interceptionContext);
        }

        public override void NonQueryExecuting(DbCommand command, DbCommandInterceptionContext<int> interceptionContext)
        {
            SimulateTransientFailure(interceptionContext);
            base.NonQueryExecuting(command, interceptionContext);
        }

        private void SimulateTransientFailure<TResult>(DbCommandInterceptionContext<TResult> context)
        {
            // Simulate a transient failure 10% of the time
            double r = _random.NextDouble();
            if (r < 0.1)
            {
                var ex = new SimulatedTransientSqlException();
                string info = "Throwing a transient SqlException. ";
                Trace.WriteLine($"{info} {ex.ToString()}");
                context.Exception = ex;
            }
        }
    }

    public class SimulatedTransientSqlException : Exception
    {
        public SimulatedTransientSqlException()
        : base("Simulated transient SQL exception.") { }
    }

}


Next up connecting the dots in the DbContext, setting up the db configuration

ApplicationDbContext.cs



[DbConfigurationType(typeof(ApplicationDbConfiguration))]
public class ApplicationDbContext : DbContext
{

    static ApplicationDbContext()
    {
        if (!AppDomain.CurrentDomain.GetAssemblies().Any(a => a.FullName.StartsWith("Effort")))
        {
            DbInterception.Add(new TransientFailureInterceptor()); //add an interceptor that simulates a transient connection error occuring (30% chance of it happening)
            DbInterception.Add(new SerilogCommandInterceptor()); //do not add logging if EF6 Effor is used (for unit testing)
        }
    }
    
    //more code..


Also note that you usually do not want to add the TransientFailureInterceptor, it is just added for testing. You could for example add a boolean property on your DbContext to set if you are testing out connection resiliency and add the TransientFailureInterceptor when you can to test it, or provide a public method to add the TransientfailureInterceptor, and remove it afterwards if desired. Within a Test-project, you should be able to test out connection resiliency.

Tuesday, 24 June 2025

Custom code conventions in Entity Framework

This article will once more look at code conventions in Entity Framework. A custom code convention will be added where if the property (column of an entity) is called "Key" and is of type Guid, it is set as the key of the entity (surrounding type the property resides in). The code in this article is available in my Github repo here:

https://github.com/toreaurstadboss/BulkOperationsEntityFramework

The custom code convention for setting all properties called Key of property type Guid as the key of an entity (table) looks like the following:

GuidKeyConvention.cs



using System;
using System.Data.Entity.ModelConfiguration.Conventions;
using System.Linq;
using System.Reflection;

namespace BulkOperationsEntityFramework.Conventions
{

    public class GuidKeyConvention : Convention
    {
        public GuidKeyConvention()
        {
            Types().Configure(t =>
            {
                var keyProperty = t.ClrType
                    .GetProperties(BindingFlags.Public | BindingFlags.Instance)
                    .FirstOrDefault(p => p.PropertyType == typeof(Guid)
                    && string.Equals(p.Name, "Key", StringComparison.OrdinalIgnoreCase));

                if (keyProperty != null)
                {
                    t.HasKey(keyProperty);
                }
            });            
        }
    }

}



The custom code convention can then be added in the OnModelCreating method shown below :

ApplicationDbContext.cs



 protected override void OnModelCreating(DbModelBuilder modelBuilder)
 {
     modelBuilder.Entity<User>().HasKey(u => u.Id);
     modelBuilder.Entity<User>().Property(u => u.Id).HasDatabaseGeneratedOption(DatabaseGeneratedOption.Identity);

     modelBuilder.Properties<string>().Configure(p => p.HasMaxLength(255)); // Set max length for all string properties

     modelBuilder.Conventions.Add(new GuidKeyConvention());
     
     //more code

 }


As shown above, a custom convention inherits from the System.Data.Entity.ModelConfiguration.Conventions.Convention class. The behavior of the custom code convention is set up in the constructor of the custom code convention. It is not overriding any methods, instead it is making use over inherited public methods Types() or Properties() or Properties(). An example of an entity that then will use this custom code convention is shown with the following entity (POCO) :

Session.cs



using System;

namespace BulkOperationsEntityFramework.Models
{
    public class Session
    {

        public Guid Key { get; set; } // Primary key by convention

        public DateTime CreatedAt { get; set; }

        public DateTime? ExpiresAt { get; set; }

        public string IpAddress { get; set; }

        public string UserAgent { get; set; }
    }
}


Adding this custom code convention, the following database migration is then added and shows that the property (field/column) called Key of type Guid.


namespace BulkOperationsEntityFramework.Migrations
{
    using System.Data.Entity.Migrations;

    public partial class Sessions : DbMigration
    {
        public override void Up()
        {
            CreateTable(
                "dbo.Sessions",
                c => new
                {
                    Key = c.Guid(nullable: false),
                    CreatedAt = c.DateTime(nullable: false),
                    ExpiresAt = c.DateTime(),
                    IpAddress = c.String(maxLength: 255),
                    UserAgent = c.String(maxLength: 255),
                })
                .PrimaryKey(t => t.Key);

        }

        public override void Down()
        {
            DropTable("dbo.Sessions");
        }
    }
}


As shown in the database migration, the primary key is set to the field Key, which means that the field (property/column) Key is set as the primary key. It is of course easier to just attribute the property Key with the Key attribute or set up the primary key in the OnModelConfiguring (Fluent API). Custom code conventions are best in use when you make a custom code convention that saves a lot of setup where you have a large data model and want to standardize code conventions, this sample just is a demonstration of how such a custom code convention can be created by inheriting the Convention class in the namespace System.Data.Entity.ModelConfiguration.Conventions. In the previous article, a custom code convention encapsulated in a custom marker attribute and wiring up the logic via helper extension methods. Instead of inheriting from the Convention class, the wiring up of Schema attribute shown in previous article could be more standardized in this way. Such a SchemaConvention could be set up like this instead, inheriting Convention class.

SchemaConvention.cs




using BulkOperationsEntityFramework.Attributes;
using System.Data.Entity.Infrastructure.Pluralization;
using System.Data.Entity.ModelConfiguration.Conventions;
using System.Reflection;

namespace BulkOperationsEntityFramework.Conventions
{
    public class SchemaConvention : Convention
    {
        public SchemaConvention()
        {
            var pluralizer = new EnglishPluralizationService();

            Types().Configure(c =>
            {
                var schemaAttr = c.ClrType.GetCustomAttribute<SchemaAttribute>(false);
                var tableName = pluralizer.Pluralize(c.ClrType.Name);

                if (schemaAttr != null && !string.IsNullOrEmpty(schemaAttr.SchemaName))
                {
                    c.ToTable(tableName, schemaAttr.SchemaName ?? "dbo");
                }
                else
                {
                    c.ToTable(tableName);
                }
            });
        }
    }
}  



Saturday, 21 June 2025

Creating a convention based Schema attribute in Entity Framework for .NET Framework

This article will present a way to create a Schema attribute for Entity Framework in .NET Framework. Some people still use .NET Framework, hopefully the latest supported version .NET Framework 4.8, due to compability reasons and legacy code and having a Schema attribute is not supported in Entity Framework for .NET Framework, still in Entity Framework 6.5.0 which is the newest version. This is similar to the way we can set Schema via attribute on an entity in Entity Framework .NET Core 8. The code presented in the article is in the following Github repo : https://github.com/toreaurstadboss/BulkOperationsEntityFramework The Schema attribute is a simple marker attribute where the schema name to be set to the entity (class) upon it is applied on.

SchemAttribute.cs



using System;

namespace BulkOperationsEntityFramework.Attributes
{

    [AttributeUsage(AttributeTargets.Class, AllowMultiple = false)]
    public class SchemaAttribute : Attribute
    {
        private readonly string _schemaName;

        public SchemaAttribute(string schemaName)
        {
            _schemaName = schemaName ?? "dbo"; //fallback to default schema 'dbo' if null is passed in here             
        }

        public string SchemaName => _schemaName;

    }

}


The attribute will be used as an attribute-based code convention for Entity Framework. First off, the code conventions are set up using a helper extension. Note that it is not necessary to cast this to DbContext here, it just makes it more readable that we are passing in DbContext here.

ApplicationDbContext.cs



protected override void OnModelCreating(DbModelBuilder modelBuilder) {

// more code here if needed

  modelBuilder.ApplyCustomCodeConventions((DbContext)this); // Apply custom code conventions based on DbSet types. Pass in db context.

// more code here if needed

}

The helper method ApplyCustomCodeConventions will loop through all the DbSet generic properties of the passed in DbContext. DbModelBuilder is the instance that the helper extension method provides more functionality on.

ModelBuilderExtensions.cs



using BulkOperationsEntityFramework.Attributes;
using System.Data.Entity;
using System.Linq;
using System.Reflection;

namespace BulkOperationsEntityFramework
{

    public static class ModelBuilderExtensions
    {

        /// <summary>
        /// Applies custom code conventions to the specified <see cref="DbModelBuilder"/> instance based on the <see
        /// cref="DbSet{TEntity}"/> types defined in the provided <see cref="DbContext"/>.
        /// </summary>
        /// <remarks>This method inspects the <see cref="DbSet{TEntity}"/> properties of the provided <see
        /// cref="DbContext"/> and applies schema conventions to each entity type. It is typically used to enforce
        /// custom schema rules or configurations during model creation.</remarks>
        /// <param name="modelBuilder">The <see cref="DbModelBuilder"/> instance to which the conventions will be applied.</param>
        /// <param name="context">The <see cref="DbContext"/> containing the <see cref="DbSet{TEntity}"/> types to analyze.</param>
        public static void ApplyCustomCodeConventions(this DbModelBuilder modelBuilder, DbContext context)
        {
            var dbSetTypes = context
                .GetType()
                .GetProperties(BindingFlags.Instance | BindingFlags.Public)
                .Where(p => p.PropertyType.IsGenericType && p.PropertyType.GetGenericTypeDefinition() == typeof(DbSet<>))
                .Select(p => p.PropertyType.GetGenericArguments()[0]);

            foreach (var type in dbSetTypes)
            {
                ApplySchemaAttributeConvention(modelBuilder, type);
            }

        }

        /// <summary>
        /// Adds a convention to apply the Schema attribute to set the schema name of entities in the DbContext.
        /// </summary>
        /// <param name="modelBuilder"></param>
        /// <param name="type"></param>
        private static void ApplySchemaAttributeConvention(DbModelBuilder modelBuilder, System.Type type)
        {
            var schema = type.GetCustomAttribute<SchemaAttribute>(false)?.SchemaName;
            if (schema != null)
            {
                var entityMethod = typeof(DbModelBuilder).GetMethod("Entity").MakeGenericMethod(type);
                var entityTypeConfiguration = entityMethod.Invoke(modelBuilder, null);
                var toTableMethod = entityTypeConfiguration.GetType().GetMethod("ToTable", new[] { typeof(string), typeof(string) });
                toTableMethod.Invoke(entityTypeConfiguration, new object[] { type.Name, schema });
            }
        }
    }

}


The schema attribute could also set one and one entity in the OnModelConfiguration method like for example the following:

ApplicationDbContext.cs



    modelBuilder.Types().Where(p => p.GetCustomAttributes(false).OfType<SchemaAttribute>().Any())
        .Configure(t => t.ToTable(t.ClrType.Name, t.ClrType.GetCustomAttribute<SchemaAttribute>().SchemaName ?? "dbo")); //add support for setting Schema via Schema attribute using custom code convention


But the code above is much better but in a reusable helper method as shown higher up in this article so you can easier just paste in the helper method and do a much cleaner call in your DbContext's OnConfiguring method. Note that the Schema attribute is available if you use .NET Core Entity Framework Core 8, that is using .NET 8. Example usage:


ArchivedUser.cs



using BulkOperationsEntityFramework.Attributes;

namespace BulkOperationsEntityFramework.Models
{

    [Schema("Archive")]
    public class ArchivedUser
    {

        public int Id { get; set; }

        public string Email { get; set; }

        public string FirstName { get; set; }

        public string LastName { get; set; }

        public string PhoneNumber { get; set; }

    }
    
}


When this is set up, we can easily set up the schema of an entity by just attributing the table using this Schema attribute.