I came across your article while reading this post. Thank you for the wonderful article!

I have made changes to ensure consistency with the current value.

Previously, the current value would continuously increase if there were no errors. However, if a problem occurred with the backend selected by GetNextPeer func, the current value would fall within the range of 0 <= current < len(s.backends).

		idx := i % len(s.backends)     // take an index by modding
		if s.backends[idx].IsAlive() { // if we have an alive backend, use it and store if its not the original one
			if i != next {
				atomic.StoreUint64(&s.current, uint64(idx)) // Forces the value to be in the range `0 <= current < 
			}
			return s.backends[idx]
		}

I believe current should not continuously increase but should rather stay within the range of 0 <= current < len(s.backends), which is why I made this change.

I hope my PR is helpful to you.