eea722e created the issue

Cyberduck allows the user to set a maximum number of auto-retries. This limit should not be applied cumulatively to the entire task; rather, it should apply to each network failure.

For example, say I've configured 9 auto-retry attempts, and during a big synchronization job the app experiences a network failure and recovers after three retries. At that point, the next failure should have the opportunity to retry 9 times, not 6. On a job as big as the one I'm trying to do, I eventually hit the 9 retry limit, even though all the previous failures recovered on auto-retry.