Cassandra Datastax 驱动程序重试策略答案

【问题标题】：Cassandra Datastax Driver Retry PolicyCassandra Datastax 驱动程序重试策略
【发布时间】：2019-02-21 21:45:58
【问题描述】：

我们可以像这样创建一个集群实例。

 cluster = Cluster
.builder()
.addContactPoint("192.168.0.30")
.withRetryPolicy(DefaultRetryPolicy.INSTANCE)
.build();

我们是否会提供在查询时必须重试请求的次数信息。

感谢任何建议。

谢谢

【问题讨论】：

标签： cassandra

【解决方案1】：

为了指定这些值，您必须创建自己的 RetryPolicy interface 实现。

以下只是一个示例——按照您的规则进行自己的实现：

public class CustomRetryPolicy implements RetryPolicy {

    private final int readAttempts;
    private final int writeAttempts;
    private final int unavailableAttempts;

    public CustomRetryPolicy(int readAttempts, int writeAttempts, int unavailableAttempts) {
        this.readAttempts = readAttempts;
        this.writeAttempts = writeAttempts;
        this.unavailableAttempts = unavailableAttempts;
    }

    @Override
    public RetryDecision onReadTimeout(Statement stmnt, ConsistencyLevel cl, int requiredResponses, int receivedResponses, boolean dataReceived, int rTime) {
        if (dataReceived) {
            return RetryDecision.ignore();
        } else if (rTime < readAttempts) {
            return RetryDecision.retry(cl);
        } else {
            return RetryDecision.rethrow();
        }

    }

    @Override
    public RetryDecision onWriteTimeout(Statement stmnt, ConsistencyLevel cl, WriteType wt, int requiredResponses, int receivedResponses, int wTime) {
        if (wTime < writeAttempts) {
            return RetryDecision.retry(cl);
        }
        return RetryDecision.rethrow();
    }

    @Override
    public RetryDecision onUnavailable(Statement stmnt, ConsistencyLevel cl, int requiredResponses, int receivedResponses, int uTime) {
                if (uTime < unavailableAttempts) {
            return RetryDecision.retry(ConsistencyLevel.ONE);
        }
        return RetryDecision.rethrow();
    }

}

这真的很容易做到......然后你将它传递给你的Cluster......

RetryPolicy rc = new CustomRetryPolicy(3, 3, 2);
Cluster cluster = Cluster.builder().addContactPoint("192.168.0.30").withRetryPolicy(rc).build();

HTH, 卡罗

【讨论】：

Cassandra 附带的重试策略的默认重试次数是多少？？
只有一次......据我所知，所有默认实现只重试一次......有一个默认实现以较低的一致性级别重试。
请考虑从您的 onReadTimeout 中删除 if(dataRecieved) { return RetryDecision.ignore(); } 语句。如果用户请求了 quorum，这将导致数据安静地通过，并且用户认为他们有 quorum，而实际上并没有。

【解决方案2】：

对于使用 CassandraCSharpDriver NuGet Package for .NET 的任何 C# 开发人员，我无耻地复制了 Carlo Bertuccini 的出色答案并转换为 C# 语法：

CustomRetryPolicy customRetryPolicy = new CustomRetryPolicy(10, 10, 10);
cluster = Cluster.Builder().WithRetryPolicy(customRetryPolicy).WithCredentials(USER, PASS).AddContactPoint(NODE).WithCompression(CompressionType.Snappy).Build();

类：

public class CustomRetryPolicy : IRetryPolicy
{
    private int readAttempts;
    private int writeAttempts;
    private int unavailableAttempts;

    public CustomRetryPolicy(int readAttempts, int writeAttempts, int unavailableAttempts)
    {
        this.readAttempts = readAttempts;
        this.writeAttempts = writeAttempts;
        this.unavailableAttempts = unavailableAttempts;
    }

    public RetryDecision OnReadTimeout(IStatement query, ConsistencyLevel cl, int requiredResponses, int receivedResponses, bool dataRetrieved, int nbRetry)
    {
        if (dataRetrieved)
        {
            return RetryDecision.Ignore();
        }
        else if (nbRetry < readAttempts)
        {
            return RetryDecision.Retry(cl);
        }
        else
        {
            return RetryDecision.Rethrow();
        }
    }

    public RetryDecision OnUnavailable(IStatement query, ConsistencyLevel cl, int requiredReplica, int aliveReplica, int nbRetry)
    {
        if (nbRetry < unavailableAttempts)
        {
            return RetryDecision.Retry(ConsistencyLevel.One);
        }
        return RetryDecision.Rethrow();
    }

    public RetryDecision OnWriteTimeout(IStatement query, ConsistencyLevel cl, string writeType, int requiredAcks, int receivedAcks, int nbRetry)
    {
        if (nbRetry < writeAttempts)
        {
            return RetryDecision.Retry(cl);
        }
        return RetryDecision.Rethrow();
    }
}

【讨论】：