Astyanax不尊重配置的套接字超时

Question

这是我的设计：

• 4个节点Cassandra 1.2.19
• Astyanax 1.56.49

我正在设置配置，如

AstyanaxContext<Keyspace> context = new AstyanaxContext.Builder()
    .forCluster(service.getClusterName())
    .forKeyspace(service.getKeySpaceName())
    .withAstyanaxConfiguration(new AstyanaxConfigurationImpl()
         .setDiscoveryType(NodeDiscoveryType.NONE)
         .setCqlVersion("3.0.0")
         .setDefaultReadConsistencyLevel(consistencyLevel.getAstyanaxValue())
         .setDefaultWriteConsistencyLevel(consistencyLevel.getAstyanaxValue())
    )
    .withConnectionPoolConfiguration(new ConnectionPoolConfigurationImpl("b2bConnectionPool")
         .setPort(service.getPort())
         .setMaxConnsPerHost(5)
         .setSeeds(StringUtils.join(hosts, ","))
         // increase default timeout for heavy operations (milliseconds)
         .setSocketTimeout(15000)
         .setSSLConnectionContext(sslContext)
         .setAuthenticationCredentials(credentials)
    )
    .withConnectionPoolMonitor(new CountingConnectionPoolMonitor())
    .buildKeyspace(ThriftFamilyFactory.getInstance());

现在有一个可重复的查询，需要很长的时间，并最终抛出一个OperationTimeoutException。

com.netflix.astyanax.connectionpool.exceptions.OperationTimeoutException: OperationTimeoutException: [host=myhost(myip):13260, latency=10001(40007), attempts=4]TimedOutException()
    at com.netflix.astyanax.thrift.ThriftConverter.ToConnectionPoolException(ThriftConverter.java:171) ~[astyanax-thrift-1.56.49.jar:na]
    at com.netflix.astyanax.thrift.AbstractOperationImpl.execute(AbstractOperationImpl.java:65) ~[astyanax-thrift-1.56.49.jar:na]
    at com.netflix.astyanax.thrift.ThriftColumnFamilyQueryImpl$1$2.execute(ThriftColumnFamilyQueryImpl.java:190) ~[astyanax-thrift-1.56.49.jar:na]
    at com.netflix.astyanax.thrift.ThriftColumnFamilyQueryImpl$1$2.execute(ThriftColumnFamilyQueryImpl.java:182) ~[astyanax-thrift-1.56.49.jar:na]
    at com.netflix.astyanax.thrift.ThriftSyncConnectionFactoryImpl$ThriftConnection.execute(ThriftSyncConnectionFactoryImpl.java:151) ~[astyanax-thrift-1.56.49.jar:na]
    at com.netflix.astyanax.connectionpool.impl.AbstractExecuteWithFailoverImpl.tryOperation(AbstractExecuteWithFailoverImpl.java:119) ~[astyanax-core-1.56.49.jar:na]
    at com.netflix.astyanax.connectionpool.impl.AbstractHostPartitionConnectionPool.executeWithFailover(AbstractHostPartitionConnectionPool.java:338) ~[astyanax-core-1.56.49.jar:na]
    at com.netflix.astyanax.thrift.ThriftColumnFamilyQueryImpl$1.execute(ThriftColumnFamilyQueryImpl.java:180) ~[astyanax-thrift-1.56.49.jar:na]

异常消息写着"latency=10001“，我认为这应该是配置为15000 ms的套接字超时，但显然不是。如何在astyanax中增加查询操作的超时？

Answer 1

当我试图使用更大的数据查询节点时，我得到了类似的超时异常。我在cassandra.yaml中修改了这4个值，它解决了所有超时错误。

协调程序应该等待多久才能完成read_request_timeout_in_ms: 15000的读操作

协调程序应等待seq或索引扫描多久才能完成range_request_timeout_in_ms: 30000

协调程序应该等待多久才能完成write_request_timeout_in_ms: 30000

协调程序应等待计数器写入多长时间才能完成counter_write_request_timeout_in_ms: 15000

注意:您需要在集群中的所有节点中执行此操作，还需要在所有节点中重新启动Cassandra。

Answer 2

我相信您没有遇到套接字超时，而是我认为不能通过Astyanax控制的rpc超时。在堆栈跟踪中，封装在OperationTimeoutException中的异常是TimedOutException，当发生rpc超时时(默认为10秒)将抛出该异常。如果出现套接字超时，将抛出SocketTimeoutException。

尝试使用cqlsh执行查询，您将得到一条Request did not complete within rpc_timeout消息。

在您的<cassandra_home>/conf/cassandra.yaml中尝试设置：

read_request_timeout_in_ms: 20000
rpc_timeout_in_ms: 20000

重新启动服务器并再次尝试执行查询。