Redis Stream简介
Redis Stream是随着5.0版本发布的一种新的Redis数据类型:
高效消费者组:允许多个消费者组从同一数据流的不同部分消费数据,每个消费者组都能独立地处理消息,这样可以并行处理和提高效率。
阻塞操作:消费者可以设置阻塞操作,这样它们会在流中有新数据添加时被唤醒并开始处理,这有助于减少资源消耗并提高响应速度。
数据持久化:它可以将数据持久化到内存(配置本地持久化后会写入到存储设备)中进行保存,等待消费。
多生产者多消费者:Redis Streams能够在多个生产者和消费者之间建立一个数据通道,使得数据的流动和处理更加灵活。
扩展性和异步通信:用户可以通过应用程序轻松扩展消费者数量,并且生产者和消费者之间的通信可以是异步的,这有助于提高系统的整体性能。
满足多样化需求:Redis Streams满足从实时数据处理到历史数据访问的各种需求,同时保持易于管理。
Redis Stream可以干什么
消息队列:Redis Stream可以用作一个可靠的消息队列系统,支持发布/订阅模式,生产者和消费者可以异步地发送和接收消息。
任务调度:Redis Stream可以用于实现分布式任务调度系统,将任务分发到多个消费者进行处理,从而提高处理速度和系统可扩展性。
事件驱动架构:Redis Stream可以作为事件驱动架构中的核心组件,用于处理来自不同服务的事件,实现解耦和灵活性。
FreeRedis简介
FreeRedis 的命名来自,“自由”、“免费”,它和名字与 FreeSql 是一个理念,简易是他们一致的追寻方向,最低可支持 .NET Framework 4.0 运行环境,支持到 Redis-server 7.2。
github MIT开源协议
作者博客园地址
官方介绍
基于 .NET 的 Redis 客户端,支持 .NET Core 2.1+、.NET Framework 4.0+、Xamarin 以及 AOT。
- 所有方法名与 redis-cli 保持一致
- 支持 Redis 集群(服务端要求 3.2 及以上版本)
- 支持 Redis 哨兵模式
- 支持主从分离(Master-Slave)
- 支持发布订阅(Pub-Sub)
- 支持 Redis Lua 脚本
- 支持管道(Pipeline)
- 支持事务
- 支持 GEO 命令(服务端要求 3.2 及以上版本)
- 支持 STREAM 类型命令(服务端要求 5.0 及以上版本)
- 支持本地缓存(Client-side-cahing,服务端要求 6.0 及以上版本)
- 支持 Redis 6 的 RESP3 协议
要实现的功能
1、生产者生产数据
2、消费者消费数据后确认
3、消费者消费数据后不确认
4、已消费但超时未确认的消息监控
5、已消费但超时未确认的消息二次消费
项目依赖
WPF
CommunityToolkit.Mvvm
FreeRedis
Newtonsoft.Json
NLog
redis-windows-7.2.5
业务场景代码
涉及到的Redis命令
创建消费者组 XGROUP CREATE key group <id | $> [MKSTREAM] [ENTRIESREAD entries-read]
查询消费者组信息 XINFO STREAM key [FULL [COUNT count]]
消息队列数量(长度) XLEN key
添加消息到队列尾部 XADD key [NOMKSTREAM] [<MAXLEN | MINID> [= | ~] threshold [LIMIT count]] <* | id> field value [field value ...]
消费组成员读取消息 XREADGROUP GROUP group consumer [COUNT count] [BLOCK milliseconds] [NOACK] STREAMS key [key ...] id [id ...]
确认消息 XACK key group id [id ...]
删除消息 XDEL key id [id ...]
获取消费未确认消息的队列信息 XPENDING key group [[IDLE min-idle-time] start end count [consumer]]
把未消费的消息消费者转移到当前消费者名下 XAUTOCLAIM key group consumer min-idle-time start [COUNT count] [JUSTID]
代码
App.xaml.cs
public partial class App : Application
{
public static Logger Logger = LogManager.GetCurrentClassLogger();
public static RedisClient RedisHelper;
public static MainViewModel MainViewModel;
private void App_OnStartup(object sender, StartupEventArgs e)
{
Current.DispatcherUnhandledException += Current_DispatcherUnhandledException;
try
{
//redis6以上版本启用了ACL用户管理机制,默认用户名是default,可以忽略密码
RedisHelper = new RedisClient("127.0.0.1:6379,user=defualt,defaultDatabase=13");
RedisHelper.Serialize = JsonConvert.SerializeObject;//序列化
RedisHelper.Deserialize = JsonConvert.DeserializeObject;//反序列化
RedisHelper.Notice += (s, ee) => Console.WriteLine(ee.Log); //打印命令日志
MainViewModel = new MainViewModel();
}
catch (Exception exception)
{
MessageBox.Show(exception.Message);
Current.Shutdown(-100);
}
}
private void Current_DispatcherUnhandledException(object sender, System.Windows.Threading.DispatcherUnhandledExceptionEventArgs e)
{
e.Handled = true;
Logger.Error($"未捕获的错误:来源:{sender},错误:{e}");
}
private void App_OnExit(object sender, ExitEventArgs e)
{
RedisHelper.Dispose();
}
}
MainWindow.xaml的主要内容
<StackPanel>
<WrapPanel ItemHeight="40" Margin="10,10,0,0" VerticalAlignment="Center">
<TextBlock Text="生产数据数:" VerticalAlignment="Center"></TextBlock>
<TextBox Text="{Binding RecordCount}" Width="100" MaxLength="9" VerticalAlignment="Center"></TextBox>
<TextBlock Text="生产者数量:" Margin="10,0,0,0" VerticalAlignment="Center"></TextBlock>
<TextBox Text="{Binding TaskCount}" Width="100" MaxLength="6" VerticalAlignment="Center"></TextBox>
<TextBlock Margin="10,0,0,0" VerticalAlignment="Center">
<Run Text="正在产生第:"></Run>
<Run Text="{Binding ProducerIndex}"></Run>
<Run Text="条数据"></Run>
</TextBlock>
<Button Content="生成数据" HorizontalAlignment="Left" Margin="10,0,0,0" VerticalAlignment="Center" Width="103" Command="{Binding ProducerRelayCommand}" />
</WrapPanel>
<WrapPanel ItemHeight="40" Margin="0,10,0,0" VerticalAlignment="Center">
<TextBlock Text="消费者数量:" Margin="10,0,0,0" VerticalAlignment="Center"></TextBlock>
<TextBox Text="{Binding ConsumerCount}" Width="100" MaxLength="6" VerticalAlignment="Center"></TextBox>
<TextBlock Margin="10,0,0,0" VerticalAlignment="Center">
<Run Text="剩余未消费:"></Run>
<Run Text="{Binding ConsumeIndex}"></Run>
<Run Text="条数据。"></Run>
</TextBlock>
<CheckBox IsChecked="{Binding IsAutoAck,Mode=TwoWay}" Content="自动确认" VerticalAlignment="Center" ></CheckBox>
<Button Content="开始消费" HorizontalAlignment="Left" Margin="10,0,0,0" VerticalAlignment="Center" Width="103" Command="{Binding ConsumeRelayCommand}" />
<Button Content="消费未确认消费队列" HorizontalAlignment="Left" Margin="10,0,0,0" VerticalAlignment="Center" Width="103" Command="{Binding PendingRelayCommand}" />
</WrapPanel>
<WrapPanel>
<StackPanel>
<TextBlock Margin="10,0,0,0" Text="队列信息:" VerticalAlignment="Center"></TextBlock>
<TextBox Margin="10,0,0,0" VerticalAlignment="Center" Height="310" Width="250" Text="{Binding StreamInfo}" VerticalScrollBarVisibility="Auto"></TextBox>
</StackPanel>
<StackPanel>
<TextBlock Margin="13,0,0,0" Text="消费信息:" VerticalAlignment="Center"></TextBlock>
<TextBox Margin="13,0,0,0" VerticalAlignment="Center" Height="310" Width="250" Text="{Binding ConsumeInfo}" VerticalScrollBarVisibility="Auto" TextWrapping="WrapWithOverflow"></TextBox>
</StackPanel>
<StackPanel>
<TextBlock Margin="13,0,0,0" Text="未确认消费信息:" VerticalAlignment="Center"></TextBlock>
<TextBox Margin="13,0,0,0" VerticalAlignment="Center" Height="310" Width="250" Text="{Binding PendingInfo}" VerticalScrollBarVisibility="Auto" TextWrapping="WrapWithOverflow"></TextBox>
</StackPanel>
<StackPanel>
<TextBlock Margin="13,0,0,0" Text="消费未确认信息:" VerticalAlignment="Center"></TextBlock>
<TextBox Margin="13,0,0,0" VerticalAlignment="Center" Height="310" Width="250" Text="{Binding PendingConsumeInfo}" VerticalScrollBarVisibility="Visible" HorizontalScrollBarVisibility="Auto" TextWrapping="WrapWithOverflow"></TextBox>
</StackPanel>
</WrapPanel>
</StackPanel>
</Grid>
MainWindow.xaml.cs
public partial class MainWindow : Window
{
public MainWindow()
{
InitializeComponent();
DataContext = App.MainViewModel;
}
private void MainWindow_OnLoaded(object sender, RoutedEventArgs e)
{
App.MainViewModel.RecordCount = 1000;
App.MainViewModel.TaskCount = 5;
App.MainViewModel.ConsumerCount = 1;
App.MainViewModel.ConsumeInfo = "等待消费信息……";
App.MainViewModel.PendingInfo = "加载中……";
App.MainViewModel.StreamInfo = "加载中……";
}
}
MainViewModel的声明和变量定义
public class MainViewModel : ObservableObject
{
#region 变量定义
private readonly string _streamKey = "redisstream";
private readonly string _consumeGroupName = "counsumeGroup";
private DateTime _utcTime = new DateTime(1970, 1, 1, 0, 0, 0);
/// <summary>
/// 生成的消息条数
/// </summary>
private static int _exchangeValue;
/// <summary>
/// 剩余未消费条数
/// </summary>
private static int _consumeValue;
/// <summary>
/// 消费信息展示队列
/// </summary>
private static ConcurrentQueue<string> _consumedQueue = new ConcurrentQueue<string>();
/// <summary>
/// 消费未确认展示队列
/// </summary>
private static ConcurrentQueue<string> _pendingConsumedQueue = new ConcurrentQueue<string>();
/// <summary>
/// 退出令牌
/// </summary>
private CancellationTokenSource _cancellationTokenSource;
/// <summary>
/// 生成消息
/// </summary>
public RelayCommand ProducerRelayCommand { get; }
/// <summary>
/// 消费消息
/// </summary>
public RelayCommand ConsumeRelayCommand { get; }
/// <summary>
/// 消费未确认信息队列消费
/// </summary>
public RelayCommand PendingRelayCommand { get; }
private int _recordCount;
/// <summary>
/// 数据条数
/// </summary>
public int RecordCount
{
get => _recordCount;
set => SetProperty(ref _recordCount, value);
}
private int _taskCount;
/// <summary>
/// 开启后台生产者数量
/// </summary>
public int TaskCount
{
get => _taskCount;
set => SetProperty(ref _taskCount, value);
}
private int _consumerCount;
/// <summary>
/// 消费者数量
/// </summary>
public int ConsumerCount
{
get => _consumerCount;
set => SetProperty(ref _consumerCount, value);
}
private int _producerIndex;
/// <summary>
/// 正在生产的序列号
/// </summary>
public int ProducerIndex
{
get => Interlocked.Exchange(ref _producerIndex, _exchangeValue);
set
{
SetProperty(ref _producerIndex, _exchangeValue);
}
}
private long _consumeIndex;
/// <summary>
/// 正在消费的序列号
/// </summary>
public long ConsumeIndex
{
get => Interlocked.Read(ref _consumeIndex);
set => SetProperty(ref _consumeIndex, value);
}
private string _streamInfo;
/// <summary>
/// 队列信息展示
/// </summary>
public string StreamInfo
{
get => _streamInfo;
set => SetProperty(ref _streamInfo, value);
}
private string _consumeInfo;
/// <summary>
/// 消费信息展示
/// </summary>
public string ConsumeInfo
{
get => _consumeInfo;
set
{
value = $"暂无消费消息[{DateTime.Now:yyyy-MM-dd HH:mm:ss}]";
if (_consumedQueue.TryDequeue(out var message))
{
value = _consumeInfo + Environment.NewLine + message;
}
SetProperty(ref _consumeInfo, value);
}
}
private string _pendingInfo;
/// <summary>
/// 消费未确认队列信息展示
/// </summary>
public string PendingInfo
{
get => _pendingInfo;
set => SetProperty(ref _pendingInfo, value);
}
private string _pedingConsumeInfo;
/// <summary>
/// 消费未确认队列的展示
/// </summary>
public string PendingConsumeInfo
{
get => _pedingConsumeInfo;
set
{
value = $"暂无未确认消费信息[{DateTime.Now:yyyy-MM-dd HH:mm:ss}]";
if (_pendingConsumedQueue.TryDequeue(out var message))
{
value = _pedingConsumeInfo + Environment.NewLine + message;
}
SetProperty(ref _pedingConsumeInfo, value);
}
}
private bool _isProduceCanExec;
/// <summary>
/// 是否可以执行生成任务
/// </summary>
public bool IsProduceCanExec
{
get => _isProduceCanExec;
set
{
SetProperty(ref _isProduceCanExec, value);
ProducerRelayCommand.NotifyCanExecuteChanged();
}
}
private bool _isAutoAck;
/// <summary>
/// 是否自动确认消费信息
/// </summary>
public bool IsAutoAck
{
get => _isAutoAck;
set
{
SetProperty(ref _isAutoAck, value);
ProducerRelayCommand.NotifyCanExecuteChanged();
}
}
#endregion
public MainViewModel()
{
ProducerRelayCommand = new RelayCommand(async () => await DoProduce(), () => !_isProduceCanExec);
ConsumeRelayCommand = new RelayCommand(async () => await DoConsume());
PendingRelayCommand = new RelayCommand(async () => await DoPendingConsume());
_cancellationTokenSource = new CancellationTokenSource();
var exist = App.RedisHelper.Exists(_streamKey);
if (!exist)
{
//创建消费组,同一个消费组可以有多个消费者,它们直接不会重复读取到同一条消息
App.RedisHelper.XGroupCreate(_streamKey, _consumeGroupName, MkStream: true);
}
else
{
var groups = App.RedisHelper.XInfoGroups(_streamKey);
if (groups == null || !groups.Any())
{
App.RedisHelper.XGroupCreate(_streamKey, _consumeGroupName);
}
}
ConsumeIndex = App.RedisHelper.XLen(_streamKey);
DoLoadStreamInfo();
DoLoadPendingInfo();
}
}
消息生成者
private async Task DoProduce()
{
if (IsProduceCanExec)
{
return;
}
IsProduceCanExec = true;
if (TaskCount > RecordCount)
{
TaskCount = RecordCount;
}
_exchangeValue = 0;
ProducerIndex = 0;
if (TaskCount > 0)
{
var pageSize = RecordCount / TaskCount;
var tasks = Enumerable.Range(1, TaskCount).Select(x =>
{
return Task.Run(() =>
{
var internalPageSize = pageSize;
if (TaskCount > 1 && x == TaskCount)
{
if (x * internalPageSize < RecordCount)
{
internalPageSize = RecordCount - (TaskCount - 1) * internalPageSize;
}
}
for (var i = 1; i <= internalPageSize; i++)
{
ProducerIndex = Interlocked.Increment(ref _exchangeValue);
ConsumeIndex = Interlocked.Increment(ref _consumeValue);
var dic = new Dictionary<string, MessageModel> { { $"user_{x}", new MessageModel { Age = 16, Description = $"描述:{ProducerIndex}", Id = 1, Name = "wang", Status = 1 } } };
App.RedisHelper.XAdd(_streamKey, 0, "*", dic);
}
return Task.CompletedTask;
});
});
await Task.WhenAll(tasks);
}
IsProduceCanExec = false;
}
#endregion
消息消费者
private Task DoConsume()
{
var groups = App.RedisHelper.XInfoGroups(_streamKey);
if (groups == null || !groups.Any())
{
App.RedisHelper.XGroupCreate(_streamKey, _consumeGroupName);
}
//添加消费者
var tasks = Enumerable.Range(1, _consumerCount).Select(c =>
{
var task = Task.Run(async () =>
{
//从消费组中读取消息,同一个组内的成员不会重复获取同一条消息。
var streamRead = App.RedisHelper.XReadGroup(_consumeGroupName, $"consumer{c}", 0, _streamKey, ">");
if (streamRead != null)
{
//取得消息
var id = streamRead.id;
var model = new Dictionary<string, MessageModel>(1)
{
{ streamRead.fieldValues[0].ToString(), JsonConvert.DeserializeObject<MessageModel>(streamRead.fieldValues[1].ToString()) }
};
_consumedQueue.Enqueue($"consumer{c}取到了消息{id},{DateTime.Now:yyyy-MM-dd HH:mm:ss}");
ConsumeInfo = "";
await Task.Delay(100);//模拟业务逻辑耗时
if (IsAutoAck)
{
//ACK
var success = App.RedisHelper.XAck(_streamKey, _consumeGroupName, id);
if (success > 0)
{
//xdel
App.RedisHelper.XDel(_streamKey, id);
_consumedQueue.Enqueue($"consumer{c}成功消费了消息{id},{DateTime.Now:yyyy-MM-dd HH:mm:ss}");
}
else
{
_consumedQueue.Enqueue($"consumer{c}的消息{id}加入了未确认队列,{DateTime.Now:yyyy-MM-dd HH:mm:ss}");
}
ConsumeInfo = "";
}
}
});
return task;
});
Task.WhenAll(tasks);
return Task.CompletedTask;
}
#endregion
未消费信息队列监控
private Task DoLoadStreamInfo()
{
Task.Factory.StartNew(async () =>
{
while (!_cancellationTokenSource.IsCancellationRequested)
{
StreamInfo = "正在查询队列信息……";
try
{
if (!App.RedisHelper.Exists(_streamKey))
{
StreamInfo = "队列尚未创建";
await Task.Delay(3000);
continue;
}
var info = App.RedisHelper.XInfoStream(_streamKey);
StreamInfo = info?.first_entry == null ? "队列数据为空。" : $"队列长度:{info.length}{Environment.NewLine}第一个编号:{info.first_entry.id}{Environment.NewLine}最后一个编号:{info.last_entry.id}{Environment.NewLine}更新时间:{DateTime.Now:yyyy-MM-dd HH:mm:ss}";
ConsumeIndex = info?.first_entry == null ? 0 : (int)info.length;
}
catch (Exception e)
{
StreamInfo = $"获取队列信息失败:{e.Message}";
}
await Task.Delay(3000);
}
}, _cancellationTokenSource.Token, TaskCreationOptions.LongRunning, TaskScheduler.Default);
return Task.CompletedTask;
}
#endregion
未消费成功(超时或业务逻辑执行失败)的消息队列消息展示
private Task DoLoadPendingInfo()
{
Task.Run(async () =>
{
while (!_cancellationTokenSource.IsCancellationRequested)
{
PendingInfo = "正在查询未确认队列……";
if (!App.RedisHelper.Exists(_streamKey))
{
PendingInfo = "队列尚未创建";
await Task.Delay(3000);
continue;
}
try
{
var info = App.RedisHelper.XPending(_streamKey, _consumeGroupName);
if (info == null || info.count == 0)
{
PendingInfo = $"暂无未确认信息。[{DateTime.Now:yyyy-MM-dd HH:mm:ss}]";
await Task.Delay(3000);
continue;
}
var infoTxt = $"未消费数量:{info.count}{Environment.NewLine}涉及{info.consumers.Length}个消费者{Environment.NewLine}最小编号:{info.minId}{Environment.NewLine}最大编号:{info.maxId}{Environment.NewLine}更新时间:[{DateTime.Now:yyyy-MM-dd HH:mm:ss}]";
PendingInfo = infoTxt;
}
catch (Exception e)
{
PendingInfo = $"获取队列信息失败:{e.Message}";
}
await Task.Delay(3000);
}
}, _cancellationTokenSource.Token);
return Task.CompletedTask;
}
#endregion
未消费成功的消息重新消费
private Task DoPendingConsume()
{
Task.Run(async () =>
{
while (!_cancellationTokenSource.IsCancellationRequested)
{
_pendingConsumedQueue.Enqueue($"正在查询未确认队列……[{DateTime.Now:yyyy-MM-dd HH:mm:ss}]");
PendingConsumeInfo = "";
if (!App.RedisHelper.Exists(_streamKey))
{
_pendingConsumedQueue.Enqueue($"队列尚未创建……[{DateTime.Now:yyyy-MM-dd HH:mm:ss}]");
PendingConsumeInfo = "";
await Task.Delay(3000);
continue;
}
try
{
//从stream队列的头部(0-0的位置)获取2条已读取时间超过2分钟且未确认的消息,修改所有者为pendingUser重新消费并确认。
var info = App.RedisHelper.XAutoClaim(_streamKey, _consumeGroupName, "pendingUser", 120000, "0-0", 2);
if (info == null || info.entries == null || info.entries.Length == 0)
{
_pendingConsumedQueue.Enqueue("未确认队列中暂无信息。[{DateTime.Now:yyyy-MM-dd HH:mm:ss}]");
await Task.Delay(3000);
continue;
}
foreach (var entry in info.entries)
{
if (entry == null) continue;
_pendingConsumedQueue.Enqueue($"未确认消费信息:{entry.id}[{DateTime.Now:yyyy-MM-dd HH:mm:ss}]");
Debug.WriteLine(JsonConvert.DeserializeObject<MessageModel>(entry.fieldValues[1].ToString()));
PendingConsumeInfo = "";
//ACK
await Task.Delay(100);//模拟业务逻辑执行时间
var success = App.RedisHelper.XAck(_streamKey, _consumeGroupName, entry.id);
if (success > 0)
{
//xdel
if (App.RedisHelper.XDel(_streamKey, entry.id) > 0)
{
_pendingConsumedQueue.Enqueue($"consumer[pendingUser]成功消费了消息{entry.id},{DateTime.Now:yyyy-MM-dd HH:mm:ss}");
}
else
{
_pendingConsumedQueue.Enqueue($"[pendingUser]删除{entry.id}[失败],{DateTime.Now:yyyy-MM-dd HH:mm:ss}");
}
}
else
{
_pendingConsumedQueue.Enqueue($"[pendingUser]消费{entry.id}[失败],{DateTime.Now:yyyy-MM-dd HH:mm:ss}");
}
}
}
catch (Exception e)
{
PendingConsumeInfo = $"获取队列信息失败:{e.Message}";
}
PendingConsumeInfo = "";
await Task.Delay(3000);
}
}, _cancellationTokenSource.Token);
return Task.CompletedTask;
}
#endregion
总结
本次我们通过Redis的Stream数据类型实现了部署简单、高性能、高可用性的消息队列,在中小型项目上可适用于需要处理数据流转的场景。
参考资料
① Redis Streams
②Redis Commands
没有回复内容