查看原文
其他

EF Core3.0+ 通过拦截器实现读写分离与SQL日志记录

DotNet 2021-09-23

推荐关注↓

前言


本文主要是讲解EF Core3.0+ 通过拦截器实现读写分离与SQL日志记录


注意拦截器只有EF Core3.0+ 支持,2.1请考虑上下文工厂的形式实现.


正文


一、通过拦截器实现读写分离


先讲一下本文实现的方式吧


SQL 通过数据库本身的功能 实现主从备份 大概原理如图:

EF Core在查询的时候通过**DbCommandInterceptor 拦截器(PS:这个功能在EF6.0+中也实现了)**来拦截对数据库的访问,从而切换主从数据库

下面直接上代码吧

首先我们创建一个类 继承DbCommandInterceptor:

 public class DbMasterSlaveCommandInterceptor : DbCommandInterceptor
    {
        private string _masterConnectionString;
        private string _slaveConnectionString;

        public DbMasterSlaveCommandInterceptor(string masterConnectionString, string slaveConnectionString)
        {
            _masterConnectionString = masterConnectionString;
            _slaveConnectionString = slaveConnectionString;
        }
}

通过构造函数传递主库连接地址与从库地址(可有多个 通过"|"分割)

添加一个随机分配从表读取连接的方法(PS:这里只是demo所以很简陋的随机,如果正式要用,应包含权重判断,定时心跳从库连接情况,请自行修改):

     /// <summary>
        /// 通过随机数分配获取多个从库
        /// </summary>
        /// <returns></returns>
        private string GetSlaveConnectionString()
        {
            var readArr = _slaveConnectionString.Split(new char[] { '|' }, StringSplitOptions.RemoveEmptyEntries);
            var resultConn = string.Empty;
            if (readArr != null && readArr.Any())
            {
                resultConn = readArr[Convert.ToInt32(Math.Floor((double)new Random().Next(0, readArr.Length)))];
            }
            return resultConn;
        }

添加判断是否主从操作连接方法:

        private void UpdateToSlave(DbCommand command)
        {
            //判断是否配置了主从分离
            if (!string.IsNullOrWhiteSpace(GetSlaveConnectionString()))//如果配置了读写分离,就进入判断
            {
                //判断是否为插入语句(EF 插入语句会通过Reader执行并查询主键),否则进入
                if (command.CommandText.ToLower().StartsWith("insert", StringComparison.InvariantCultureIgnoreCase) == false)
                {
                    // 判断当前会话是否处于分布式事务中
                    bool isDistributedTran = Transaction.Current != null &&
                                             Transaction.Current.TransactionInformation.Status !=
                                             TransactionStatus.Committed;
                    //判断该 context 是否处于普通数据库事务中
                    bool isDbTran = command.Transaction != null;
                    //如果不处于事务中,则执行从服务器查询
                    if (!isDbTran && !isDistributedTran)
                    {
                        command.Connection.Close();
                        command.Connection.ConnectionString = GetSlaveConnectionString();
                        command.Connection.Open();

                    }

                }
            }
        }

重载DbCommandInterceptor当中的拦截方法,代码如下:

        //如果是写入,则正常执行
        public override InterceptionResult<int> NonQueryExecuting(DbCommand command, CommandEventData eventData, InterceptionResult<int> result)
        {
            return base.NonQueryExecuting(command, eventData, result);
        }
        public override Task<InterceptionResult<int>> NonQueryExecutingAsync(DbCommand command, CommandEventData eventData, InterceptionResult<int> result, CancellationToken cancellationToken = default)
        {
            return base.NonQueryExecutingAsync(command, eventData, result, cancellationToken);
        }

        public override InterceptionResult<DbDataReader> ReaderExecuting(DbCommand command, CommandEventData eventData, InterceptionResult<DbDataReader> result)
        {
            this.UpdateToSlave(command);
            return base.ReaderExecuting(command, eventData, result);
        }

        public override Task<InterceptionResult<DbDataReader>> ReaderExecutingAsync(DbCommand command, CommandEventData eventData, InterceptionResult<DbDataReader> result, CancellationToken cancellationToken = default)
        {
            this.UpdateToSlave(command);
            return base.ReaderExecutingAsync(command, eventData, result, cancellationToken);
        }


        public override InterceptionResult<object> ScalarExecuting(DbCommand command, CommandEventData eventData, InterceptionResult<object> result)
        {
            this.UpdateToSlave(command);
            return base.ScalarExecuting(command, eventData, result);
        }

        public override Task<InterceptionResult<object>> ScalarExecutingAsync(DbCommand command, CommandEventData eventData, InterceptionResult<object> result, CancellationToken cancellationToken = default)
        {
            this.UpdateToSlave(command);
            return base.ScalarExecutingAsync(command, eventData, result, cancellationToken);
        }

最后在EF core的上下文中注入拦截器(PS:我这里使用的Autofac模块注入):

                builder.Register(
                c =>
                {
                    var optionsBuilder = new DbContextOptionsBuilder<TestEFContext>();
                   //注入拦截器
                    optionsBuilder.AddInterceptors(new DbMasterSlaveCommandInterceptor(WriteConnect, ReadConnect));
                    //MaxBatchSize 处理批量操作BUG
                    optionsBuilder.UseMysql(WriteConnect, b=>b.MaxBatchSize(1));
                    return optionsBuilder.Options;
                }
                ).As<DbContextOptions<TestEFContex>>().SingleInstance();

这样就实现了通过拦截器实现读写分离.

二、通过拦截器实现SQL日志记录

同理,我们可以通过拦截器实现EF Core SQL语句的记录与调试

首先我们创建一个新的拦截器DBlogCommandInterceptor 如下:

public class DBlogCommandInterceptor : DbCommandInterceptor
    {
    //创建一个队列记录SQL执行时间
        static readonly ConcurrentDictionary<DbCommand, DateTime> MStartTime = new ConcurrentDictionary<DbCommand, DateTime>();
        private ILogger<DBlogCommandInterceptor> _logger { get; set; }
    //通过构造函数注入日志
        public DBlogCommandInterceptor(ILogger<DBlogCommandInterceptor> Logger)
        {
            _logger = Logger;
        }
}

创建2个私有的方法,一个记录执行开始时间,一个记录SQL

      //记录SQL开始执行的时间 
        private void OnStart(DbCommand command)
        {
            MStartTime.TryAdd(command, DateTime.Now);
        }
    //通过_logger输出日志
        private void Log(DbCommand command)
        {

            DateTime startTime;
            TimeSpan duration;
            //得到此command的开始时间
            MStartTime.TryRemove(command, out startTime);
            if (startTime != default(DateTime))
            {
                duration = DateTime.Now - startTime;
            }
            else
            {
                duration = TimeSpan.Zero;
            }
            var parameters = new StringBuilder();
            //循环获取执行语句的参数值
            foreach (DbParameter param in command.Parameters)
            {
                parameters.AppendLine(param.ParameterName + " " + param.DbType + " = " + param.Value);
            }
            _logger.LogInformation("{starttime}开始执行SQL语句:{sql},参数:{canshu},执行时间{readtime}",
               startTime.ToString(), command.CommandText, parameters.ToString(), duration.TotalSeconds);

        }

最后重载拦截器的方法:

public override InterceptionResult<int> NonQueryExecuting(DbCommand command, CommandEventData eventData, InterceptionResult<int> result)
        {
            OnStart(command);
            return base.NonQueryExecuting(command, eventData, result);
        }
        public override Task<InterceptionResult<int>> NonQueryExecutingAsync(DbCommand command, CommandEventData eventData, InterceptionResult<int> result, CancellationToken cancellationToken = default)
        {
            OnStart(command);
            return base.NonQueryExecutingAsync(command, eventData, result, cancellationToken);
        }
        public override int NonQueryExecuted(DbCommand command, CommandExecutedEventData eventData, int result)
        {
            Log(command);
            return base.NonQueryExecuted(command, eventData, result);
        }
        public override Task<int> NonQueryExecutedAsync(DbCommand command, CommandExecutedEventData eventData, int result, CancellationToken cancellationToken = default)
        {
            Log(command);
            return base.NonQueryExecutedAsync(command, eventData, result, cancellationToken);
        }

        public override InterceptionResult<object> ScalarExecuting(DbCommand command, CommandEventData eventData, InterceptionResult<object> result)
        {
            OnStart(command);
            return base.ScalarExecuting(command, eventData, result);
        }
        public override Task<InterceptionResult<object>> ScalarExecutingAsync(DbCommand command, CommandEventData eventData, InterceptionResult<object> result, CancellationToken cancellationToken = default)
        {
            OnStart(command);
            return base.ScalarExecutingAsync(command, eventData, result, cancellationToken);
        }
        public override object ScalarExecuted(DbCommand command, CommandExecutedEventData eventData, object result)
        {
            Log(command);
            return base.ScalarExecuted(command, eventData, result);
        }
        public override Task<object> ScalarExecutedAsync(DbCommand command, CommandExecutedEventData eventData, object result, CancellationToken cancellationToken = default)
        {
            Log(command);
            return base.ScalarExecutedAsync(command, eventData, result, cancellationToken);
        }


        public override InterceptionResult<DbDataReader> ReaderExecuting(DbCommand command, CommandEventData eventData, InterceptionResult<DbDataReader> result)
        {
            OnStart(command);
            return base.ReaderExecuting(command, eventData, result);
        }
        public override Task<InterceptionResult<DbDataReader>> ReaderExecutingAsync(DbCommand command, CommandEventData eventData, InterceptionResult<DbDataReader> result, CancellationToken cancellationToken = default)
        {
            OnStart(command);
            return base.ReaderExecutingAsync(command, eventData, result, cancellationToken);
        }

        public override Task<DbDataReader> ReaderExecutedAsync(DbCommand command, CommandExecutedEventData eventData, DbDataReader result, CancellationToken cancellationToken = default)
        {
            Log(command);
            return base.ReaderExecutedAsync(command, eventData, result, cancellationToken);
        }
        public override DbDataReader ReaderExecuted(DbCommand command, CommandExecutedEventData eventData, DbDataReader result)
        {
            Log(command);
            return base.ReaderExecuted(command, eventData, result);
        }

这样,我们就实现了通过拦截器实现SQL日志记录~效果如下:

调试SQL语句就方便了很多~

转自:GuZhenYin

链接:cnblogs.com/GuZhenYin/p/14544714.html

- EOF -

推荐阅读  点击标题可跳转
如何在 C#9 中使用 static 匿名函数.NET 中使用 Flurl 高效处理Http请求快速了解C# 8.0 中 "可空引用类型" 语言特性


看完本文有收获?请转发分享给更多人

推荐关注「DotNet」,提升.Net技能 

点赞和在看就是最大的支持❤️

: . Video Mini Program Like ,轻点两下取消赞 Wow ,轻点两下取消在看

您可能也对以下帖子感兴趣

文章有问题?点此查看未经处理的缓存