#4 retry/timeout does not work as expected in tableretriever

open
nobody
None
5
2010-10-08
2010-10-08
No

The retry/timeout strategy in tableretriever.py does not seem to work as expected. Given the defaults of 4 retries and a 2.0 second initial timeout, I would expect 5 requests to be made before giving up entirely, and I expect the timeout to increase by 1.5 (as the doc says) for each retry; that would amount to about 26 seconds of waiting time before giving up entirely on an unresponsive host.

What happens is that only three messages are sent, and the timeout stays constant between retries.

I'm attaching a patch with tests for the apparent problems, and a two-liner fix for them.

Discussion