Nope - I also get this problem. Always accompanied by an error 1030 "got error 127 from table handler" during a thread update.
If I run check or repair on the thread table it's always fine. All the post records are still there so provided you know the threadid (provided in the error msg) and the forumid (easy to guess - or just pick one and move it later) you can always reconstruct the thread record. I was doing this a couple of times a week with phpMyAdmin but then got bored and wrote a PHP script to automate the process. Takes a threadid and forumid as arguments and reconstructs the thread record.
Still no idea what's causing the problem in the first place though
Michael.