A few weeks ago, I cut over my blog and several other Web sites to a new, much fast server. I don't know that it's made much difference in how fast people retrieve my blog since it's mostly static, but it's made a great deal of difference to me in posting speed and other back office functions.

What's been curious to me is that the old server continues to get a few hits. I did a little exploring today and discovered a few interesting things.

First, all of the hits are for RSS feeds of one kind of another.

Second, all of the clients still hitting the old server are implemented in Java. Here's a list (from the user-agent data in the log): Rome, Jakarta Commons-HttpClient, Java/1.5.0_06, and knLiveJava/3.0.

Can someone explain this to me? Is there a bug in some Java library that caches IP addresses for too long or is this just coincidence?