如何在失败后恢复对队列的写入

Question

磁盘空间完成后，我得到了InternalError。添加磁盘空间不是修复问题。

可以恢复并继续保持下去吗？

可能出错了，我可以尝试重新创建/关闭？

创建queue queue = SingleChronicleQueueBuilder.binary(basePath) .build();

单线程“TradeReactorEventPersist-1”

    ExcerptAppender appender = acquireAppender;
    if (appender == null) {
        appender = queue.acquireAppender();
        acquireAppender = appender;
    }
    appender.writeBytes(BytesStore.wrap(b));

在接下来的例外之后：

2019-08-23 08:13:26.963 +0000 ERROR [TradeReactorEventPersister-1] LoggingUncaughtExceptionHandler - Uncaught exception a fault occurred in a recent unsafe memory access operation in compiled Java code in thread TradeReactorEventPersister-1 
java.lang.InternalError: a fault occurred in a recent unsafe memory access operation in compiled Java code
        at net.openhft.chronicle.wire.AbstractWire.updateHeaderAssertions(AbstractWire.java:546)
        at net.openhft.chronicle.wire.AbstractWire.updateHeader(AbstractWire.java:533)
        at net.openhft.chronicle.queue.impl.single.SingleChronicleQueueExcerpts$StoreAppender.writeBytes(SingleChronicleQueueExcerpts.java:470)

2019-08-23 08:13:26.965 +0000 ERROR [TradeReactorEventPersister-1] LoggingUncaughtExceptionHandler - Uncaught exception a fault occurred in a recent unsafe memory access operation in compiled Java code in thread TradeReactorEventPersister-1 
java.lang.InternalError: a fault occurred in a recent unsafe memory access operation in compiled Java code
        at net.openhft.chronicle.wire.AbstractWire.updateHeaderAssertions(AbstractWire.java:547)
        at net.openhft.chronicle.wire.AbstractWire.updateHeader(AbstractWire.java:533)
        at net.openhft.chronicle.queue.impl.single.SingleChronicleQueueExcerpts$StoreAppender.writeBytes(SingleChronicleQueueExcerpts.java:470)

2019-08-23 08:13:27.166 +0000 ERROR [TradeReactorEventPersister-1] LoggingUncaughtExceptionHandler - Uncaught exception a fault occurred in a recent unsafe memory access operation in compiled Java code in thread TradeReactorEventPersister-1 
java.lang.InternalError: a fault occurred in a recent unsafe memory access operation in compiled Java code
        at net.openhft.chronicle.wire.AbstractWire.updateHeader(AbstractWire.java:511)
        at net.openhft.chronicle.queue.impl.single.SingleChronicleQueueExcerpts$StoreAppender.writeBytes(SingleChronicleQueueExcerpts.java:470)

2019-08-23 08:13:27.167 +0000 ERROR [TradeReactorEventPersister-1] LoggingUncaughtExceptionHandler - Uncaught exception you cant put a header inside a header, check that you have not nested the documents. If you are using Chronicle-Queue please ensure that you have a unique instance of the Appender per thread, in other-words you can not share appenders across threads. in thread TradeReactorEventPersister-1 
java.lang.AssertionError: you cant put a header inside a header, check that you have not nested the documents. If you are using Chronicle-Queue please ensure that you have a unique instance of the Appender per thread, in other-words you can not share appenders across threads.
        at net.openhft.chronicle.wire.AbstractWire.enterHeader(AbstractWire.java:322)
        at net.openhft.chronicle.queue.impl.single.SingleChronicleQueueExcerpts$StoreAppender.writeHeader(SingleChronicleQueueExcerpts.java:405)
        at net.openhft.chronicle.queue.impl.single.SingleChronicleQueueExcerpts$StoreAppender.writeBytes(SingleChronicleQueueExcerpts.java:463)

添加磁盘空间后，我无法坚持。每次尝试持久化事件，我都得到了最后一个例外：

2019-08-23 08:22:50.746 +0000 ERROR [TradeReactorEventPersister-1] LoggingUncaughtExceptionHandler - Uncaught exception you cant put a header inside a header, check that you have not nested the documents. If you are using Chronicle-Queue please ensure that you have a unique instance of the Appender per thread, in other-words you can not share appenders across threads. in thread TradeReactorEventPersister-1 
java.lang.AssertionError: you cant put a header inside a header, check that you have not nested the documents. If you are using Chronicle-Queue please ensure that you have a unique instance of the Appender per thread, in other-words you can not share appenders across threads.
        at net.openhft.chronicle.wire.AbstractWire.enterHeader(AbstractWire.java:322)
        at net.openhft.chronicle.queue.impl.single.SingleChronicleQueueExcerpts$StoreAppender.writeHeader(SingleChronicleQueueExcerpts.java:405)
        at net.openhft.chronicle.queue.impl.single.SingleChronicleQueueExcerpts$StoreAppender.writeBytes(SingleChronicleQueueExcerpts.java:463)

Answer 1

不幸的是，编年史队列在相当低的级别上工作，并且在数据损坏后无法自动修复自己(磁盘空间不可避免地会导致数据损坏)。顺便说一句，为了避免这种情况，如果磁盘空间不足，纪事队列会警告您，您应该看到一条警告消息，如："your disk " + this.fileStore + " is almost full, warning: chronicle-queue may crash if it runs out of space."

PS你不应该真的需要做懒散的获取逻辑。您总是可以简单地执行queue.acquireAppender() --这是一个廉价的调用，可以从ThreadLocal池中获取预先存在的附录，而且每次而不是都会创建新的附录。