Reachability test fails sometimes

Since version 0.2.4.x-alpha (I guess the whole branch up to 0.2.4.16-rc) I saw the reachability test fail on my bridge which, as far as I know, didn't show this message with 0.2.3.x.

I can't reproduce it, but it happened yesterday and the day before that, without me noticing. It runs on Windows, like always and is up part-time. Today it took extremely long (compared to previous times) to validate that it is up.

It happened before and all I did back then was to exit Vidalia (and Tor) and start it again.

Strange enough is that Tor retries the test, but keeps failing if it failed the first one. To what I have experienced it appears that this does not happen in you restart it, though I can't say that for sure.

Another bridge that was set-up during the 0.2.4.x-alpha phase on a Linux server failed this reachability test at least once and all I did was restart the Tor daemon to make the test succeed.

Both bridges are obfsproxy bridges, which are configured like normal bridges with lines for the transport. No NoAdvertise or no NoListen. There's nothing special about them.