hmm strange one - but at least its sorted.

The indexes for stuff like months are not obtained at real time - for performance reasons because this information is known when you save a document the html is generated at that point. This method helps the websites to be as quick as they are. You dont want to be doing things like counting up entries for each month, each category, each comment count etc for every page request - waste of resource.

In addition because you can have different replicas on different servers with different content because they have not replicated yet - each replica has its own store of this data. Sounds like your got temporarily confused.