I have had the same problem after upgrading to 3.2.5 from 3.2.3. However, I'm pretty sure we MAY have had the problem prior, because I saw a lot of entries in the auto-discovery portlet prior. But other users were adding servers, so it was hard for me to track.
In fact, with one agent install in a RHEL 5.2 Xen guest (post upgrade), it would always get stuck in the queue and I went through several cycles of deleting its data directory, and removing it from the EAM_AIQ tables manually as described. I'm not even sure if it started working, but I don't want to VPN to check atm 😉
I can provide more information if necessary; I mostly wanted to chime in with "me too" 🙂