为什么我的执行器服务线程挂在 java 中？

Question

I have a class which contains a scheduled executor service, to run a task every 5 seconds. Sometimes, the task itself may take longer than 5 seconds due to retry logic that I have implemented. In this case, I ASSUME that even though I have the scheduler to run the task every 5 seconds, it simply waits until the retry is complete. But, after 2 failed attempts, I see my failure log and I no longer see a call being made every 5 seconds.

import io.dropwizard.lifecycle.Managed;
import java.util.concurrent.Executors;
import java.util.concurrent.ScheduledExecutorService;
import java.util.concurrent.TimeUnit;
import lombok.NonNull;
import lombok.extern.slf4j.Slf4j;
import net.jodah.failsafe.Failsafe;
import net.jodah.failsafe.RetryPolicy;
import org.skife.jdbi.v2.Handle;
import org.skife.jdbi.v2.IDBI;

@Slf4j
public class DBTask implements Managed
{
    private final IDBI dbi;
    private final String validationQuery;
    private final ScheduledExecutorService runner;
    private final RetryPolicy policy;

    public DBHealthCheckTask(@NonNull final IDBI dbi, @NonNull final String validationQuery)
    {
        this.dbi = dbi;
        this.validationQuery = validationQuery;
        runner = Executors.newSingleThreadScheduledExecutor();
        policy = new RetryPolicy()
            .withDelay(5 , TimeUnit.SECONDS)
            .withMaxRetries(2)
            .retryOn(Exception.class);
    }

    @Override
    public void start() throws Exception
    {
        int intervalSecs = 5;
        runner.scheduleAtFixedRate(this::runWithRetry, intervalSecs, intervalSecs, TimeUnit.SECONDS);
    }

    @Override
    public void stop()
    {
        runner.shutdown();
    }

    private void runWithRetry()
    {
        Failsafe.with(policy)
            .onRetry(r -> log.info("Database connection failed.", r))
            .onFailure(r -> log.warn("WARNING- Database is unhealthy.", r))
            .run(this::run);
    }

    private void run()
    {
        try (Handle handle = dbi.open())
        {
            handle.execute(validationQuery);
        }
    }
}

Once I see the "WARNING" log, I expect that the task continues again, and then it will retry another 2 times with a delay of 5 seconds before I see another "WARNING" log. But I only see one log, after that nothing ever happens.

Answer 1

Method runWithRetry will throw exception when 2 times retry failed. For schedule, if exception caught when executing task, would not set next task execute. So if we want task always scheduled, need catch exception for task manually. In your case, do something like:

private void runWithRetry()
{
    try{
        Failsafe.with(policy)
        .onRetry(r -> log.info("Database connection failed.", r))
        .onFailure(r -> log.warn("WARNING- Database is unhealthy.", r))
        .run(this::run);
    }catch(Exception e){
        //handle exception, make sure no exception rethrow
    }
}

Answer 2

WARNING- Database is unhealthy , only printed after the last round of retry, so to add some log in run() , you'll see the retry process.

private void run() {
    // add some log here
    // System.out.println("hello");
    try (Handle handle = dbi.open()) {
        handle.execute(validationQuery);
    }
}

The explaination from JavaDoc

If multiple policies, are configured, this handler is called when execution is complete and any policy fails.