Fixed StoreTailer.toEnd() returning NOT_REACHED when called as the queue rolled

[RobAustin]

The following exception was found by a client

java.lang.IllegalStateException: NOT_REACHED index: 6c52300000000
at net.openhft.chronicle.queue.impl.single.StoreTailer.originalToEnd(StoreTailer.java:866)
at net.openhft.chronicle.queue.impl.single.StoreTailer.toEnd(StoreTailer.java:761)
at XXX.chronicle.queue.EventQueueForwarder.readWriteBytes(EventQueueForwarder.java:145)
at XXX.chronicle.queue.EventQueueForwarder.lambda$startWriteBytes$1(EventQueueForwarder.java:114)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
at java.lang.Thread.run(Thread.java:748)
at net.openhft.chronicle.core.threads.CleaningThread.run(CleaningThread.java:38)

The loop where this is thrown looks like:

while (running) {
    final long endIndex = assertionTailer.toEnd().index(); // Exception thrown here
    try (DocumentContext inDC = readTailer.readingDocument()) {
        if (inDC.isPresent()) {
            // Do something
        } else {
            // Verify that the assertion tailer index == readTailer.index()
        }
    }
}

This happened at 7pm on the dot last night - we roll the files hourly so could this have had something to do with it?

[RobAustin]

I wrote the following test case

package net.openhft.chronicle;

import net.openhft.chronicle.core.time.TimeProvider;
import net.openhft.chronicle.queue.ChronicleQueueTestBase;
import net.openhft.chronicle.queue.ExcerptAppender;
import net.openhft.chronicle.queue.ExcerptTailer;
import net.openhft.chronicle.queue.RollCycles;
import net.openhft.chronicle.queue.impl.single.SingleChronicleQueue;
import org.junit.Test;

import java.util.concurrent.Executors;
import java.util.concurrent.locks.LockSupport;

import static net.openhft.chronicle.queue.impl.single.SingleChronicleQueueBuilder.binary;

public class TestCallingToEndOnRoll extends ChronicleQueueTestBase implements TimeProvider {

    private long currentTime = 0;
    private ExcerptAppender appender;
    private ExcerptTailer tailer;

    @Test
    public void test() {
        SingleChronicleQueue build = binary(getTmpDir()).rollCycle(RollCycles.TEST_SECONDLY).timeProvider(this).build();
        appender = build.acquireAppender();

        tailer = build.createTailer();
        Executors.newSingleThreadExecutor().submit(this::append);

       // Executors.newSingleThreadExecutor().submit(this::toEnd);
        //
        LockSupport.park();
    }

    private void append() {
        for (; ; ) {
         toEnd0();
            appender.writeText("hello world");
            toEnd0();
        }

    }

    private void toEnd() {
        for (; ; ) {
            toEnd0();
        }
    }

    private void toEnd0() {
        try {
            long index = tailer.toEnd().index();
            System.out.println("index = " + index);
        } catch (IllegalStateException e) {
            e.printStackTrace();
            System.exit(-1);
        }
    }

    @Override
    public long currentTimeMillis() {
        return (currentTime = currentTime + 1000);
    }
}

and was able to reproduce this issue after running it for 38 mins

[hft-team-city]

Released in Chronicle-Queue-5.20.4, BOM-2.20.22