Wrong bootstrapping when retrying

[pat-karakun]

It seems to me that there is an off by 1 error when bootstrapping the retry mechanism (see for example here ) .
By calling AtomicInteger#incrementAndGet() instead of #getAndIncrement() the int iteration starts with 1 instead of 0 which has (at least) these two consequences (ignoring jitter):

1. there are not numRetries attempts to retry but only numRetries - 1
2. the first retry attempt is done after 2*firstBackoff (= firstBackoff * 2^1) instead of after firstBackoff (= firstBackoff * 2^0)

The two effects almost compensate each other wrt when the last attempt is roughly performed, so it's certainly not an issue for most cases.

[cescoffier]

Thanks and very nice catch! I just opened a PR fixing the issue.

[jponge]

Thanks @pat-karakun