PubSubHubbub and Polling
This blog, as I've written before, is a Jekyll install. I've done some customizations to it to allow pubsub publishing and automated tweeting; why then are the pubsub hubs still polling my server?
Currently, as a test, I'm publishing to 3 different pubsubhubbub hubs: the reference hub, SuperFeedr, and Guzzle. I'm doing this for both my atom feed and my rss feed. The idea is excellent. Instead of a service coming to my webserver and asking for the rss.xml or atom.xml files, a Jekyll plugin sends the "publish" event to the hubs. This tells the hubs that there's new content, and they should come and scrape it.
Since November 25, the Google AppEngine hosted hub has twice requested my feeds at some point after my publish event. In both instances, the secondary requests happened less than 20 seconds from the initial request.
However, SuperFeedr and Guzzle both continue to poll my server, checking for changes. Guzzle has requested the files 223 times, after only one publish event. SuperFeedr takes the cake though, polling every 12 minutes, 3913 times since November 25.
I don't want to complain about free services that help the Internet as a whole. But I guess I am. It's a waste of my bandwidth, and theirs.