Closed mattwigway closed 9 years ago
Keep track of the response times for the last, say, 100 requests, and set the chunk size based on that.
Actually we don't need to do that. We expand the chunk size each time there is an underrun. We can make it smaller each time the worker is polled and there are more than chunksize requests remaining.
Keep track of the response times for the last, say, 100 requests, and set the chunk size based on that.