Initial wait time in an exponential retry strategy.

[jcranston]

Is there a way to initialize the wait strategy to be a fixed amount of time? As in, all successive exponential wait times are added onto a fixed wait time. For example, the computeSleepTime for the exponential wait strategy which is currently

@Override
public long computeSleepTime(Attempt failedAttempt) {
    double exp = Math.pow(2, failedAttempt.getAttemptNumber());
    long result = Math.round(multiplier * exp);
    if (result > maximumWait) {
        result = maximumWait;
    }
    return result >= 0L ? result : 0L;
}

can return something like initialWait + (result >= 0L ? result : 0L)?

Perhaps this can already be accomplished in another way?

[rhuffman]

I think the closest option is this method in WaitStrategies:

    public static WaitStrategy exponentialWait(long multiplier,
                                               long maximumTime,
                                               @Nonnull TimeUnit maximumTimeUnit)

The exponential wait is normally in milliseconds. If you know your operation typically takes, say, 800 ms, then you would do something like this:

WaitStrategy = exponentialWait(1000, 5, TimeUnit.MINUTES);

That way your waits would be in seconds instead of milliseconds.

[jcranston]

Ah, interesting. Upon digging, I'm assuming it would also suffice to use the CompositeWaitStrategy as well, with a ExponentialWaitStrategy and FixedWaitStrategy.

[valera7979]

It is inconvenient. It is logical to use fixed start time and maximum time. For example, when I have a disconnection issue I want to retry after 5 seconds and exponentially increase every next attempt. Why do I need to calculate time to wait?