[Simon Lloyd]

You'll probably find that there is (or has been) an issue with the post table, when a repair was attempted it failed and left an unfinished temp file behind which you need to delete prior to repairing again then all will be good!, you'll need to use ssh to find the postindex.TMD file and delete it, when a file is attempting to repair they make another temporary file, if the databse crashes for some reason when this is taking place it leaves that file unfinished, trying to repair again does nothing as that file still exists, i suggest you backup your postindex table as it is then via ssh find that file, delete it, and repair again, you could run the Check table or analyse table first, when you see the full text it will give you the name of the file that already exists.