I see that incremental scans of repositories are now multi-threaded, so a long delay on one repository doesn't affect another. However, it looks like the initial startup scan is still single-threaded. This prevents any changes in the other repositories from being recognized until the startup scan finishes.

One of our SVN repositories just decided to start using branches/tags/trunks, so I set that up and had to reindex the whole repository... at some point, FishEye had a file lock error and the SVN repository paused itself (and nothing could fix it), so I had to restart FishEye. It's been stuck rescanning the new repository for half a day now (and will likely go on for the rest of the day), preventing the other repositories from working.