防止Ignite客户端在连接失败时日志异常和线程转储

Question

    IgniteConfiguration cfg = new IgniteConfiguration();
    cfg.setClientMode(true);
    cfg.setIgniteInstanceName(getInstanceName());
    cfg.setPeerClassLoadingEnabled(true);

    TcpDiscoveryMulticastIpFinder ipFinder = new TcpDiscoveryMulticastIpFinder();
    ipFinder.setAddresses(getIpFinderAddresses());
    cfg.setDiscoverySpi(new TcpDiscoverySpi().setIpFinder(ipFinder));

    return Ignition.start(cfg);

2021-10-05 10:34:50.781 ERROR 5548 --- [alance-service%] o.apache.ignite.internal.util.typedef.G  : Blocked system-critical thread has been detected. This can lead to cluster-wide undefined behaviour [workerName=tcp-client-disco-msg-worker, threadName=tcp-client-disco-msg-worker-#4%my-service%-#54%my-service%, blockedFor=14s]
2021-10-05 10:34:50.791  WARN 5548 --- [alance-service%]                                          : Possible failure suppressed accordingly to a configured handler [hnd=NoOpFailureHandler [super=AbstractFailureHandler [ignoredFailureTypes=UnmodifiableSet [SYSTEM_WORKER_BLOCKED, SYSTEM_CRITICAL_OPERATION_TIMEOUT]]], failureCtx=FailureContext [type=SYSTEM_WORKER_BLOCKED, err=class o.a.i.IgniteException: GridWorker [name=tcp-client-disco-msg-worker, igniteInstanceName=my-service, finished=false, heartbeatTs=1633422875964]]]
org.apache.ignite.IgniteException: GridWorker [name=tcp-client-disco-msg-worker, igniteInstanceName=my-service, finished=false, heartbeatTs=1633422875964]
  at java.base@11.0.11/java.net.PlainSocketImpl.waitForConnect(Native Method) ~[na:na]
  at java.base@11.0.11/java.net.PlainSocketImpl.socketConnect(PlainSocketImpl.java:107) ~[na:na]
  ...
2021-10-05 10:34:50.792  WARN 5548 --- [alance-service%] o.a.i.i.p.failure.FailureProcessor       : No deadlocked threads detected.
2021-10-05 10:34:50.868  WARN 5548 --- [alance-service%] o.a.i.i.p.failure.FailureProcessor       : Thread dump at 2021/10/05 10:34:50 CEST
  ...
  ... hundreds of lines with thread dump (why one would want to see them if - as stated above - there were no deadlocks?) ...
  ...
2021-10-05 10:34:57.436  WARN 5548 --- [alance-service%] o.a.i.spi.discovery.tcp.TcpDiscoverySpi  : Failed to connect to any address from IP finder (will retry to join topology every 2000 ms; change 'reconnectDelay' to configure the frequency of retries): [/127.0.0.1:47500, /127.0.0.1:47501, /127.0.0.1:47502, /127.0.0.1:47503, /127.0.0.1:47504, /127.0.0.1:47505, /127.0.0.1:47506, /127.0.0.1:47507, /127.0.0.1:47508, /127.0.0.1:47509]

Answer 1

我相信在您的例子中，最冗长的消息来自failureprocessor。当检测到故障时，它会打印一个完整的线程转储。有两个系统属性控制此行为。

1. ignite_dump_threads_on_failure-完全禁用线程转储，最好将其保留为true，因为在其他失败类型的情况下它会非常有用。
2. ignite_dump_threads_on_failure_throttling_timeout-它设置两条消息之间的最小间隔（以毫升为单位）。

下面是Javadocs:IGNITE_DUMP_THREADS_ON_FAILURE和ignite_dump_threads_on_failure_throttling_timeout。