The secret was written in a post that I linked to in my very first post about this problem. I can’t believe that I didn’t pay it any attention before.

Basically, all you need to do is to run “mythfilldatabase” without any parameters. As well as retrieving new data, it resets the database.

Now this, actually, makes sense to me. I guess the reason that I never encountered this before is that I always used to use the Radio Times grabber for downloading my programme listings, however the other year I switched to EPG (electronic programme guide). That means that I don’t run mythfilldatabase on a regular schedule.

And therefore, that means that I’m not downloading data, and cleansing the database, as much as I should be! I seem to remember something about a few thousand items being marked for something, as it scrolled up the screen. I subsequently ran it again just to check, and this time it showed zero for all items.

I since tested it, and I was able to delete around 10 different programmes, and still didn’t receive any lag in deletions. That’s going to way in around the 15 to 20GB range. That’s pretty damn good, if you ask me.

All I need to make sure is that mythfilldatabase didn’t cock anything up, and then somehow schedule it to run frequently. More than once a year I guess would be a bonus….

To make matters worse, I now get a time out on a lot of different interactions with the box, for example when I escape out of watching a tv programme. The screen will go black and the two minutes later I see a popup telling me the connection to the master backend has ‘gone away’.

I have now done an optimzation on the database through several methods and it does not appear to make life any better. It does appear to make a difference for a while but it soon returns to normal behaviour. Admittedly though I was planning on doing an ‘OPTIMIZE TABLE’ for every table in the database and so far I have only done it on the ‘program’ table.

One tool I did use was ‘myisamcheck –analyze –check-only-changed –quick –sort-index *.MYI’ in /var/lib/mysql/mythconverg

That seemed to spend sometime checking several tables, including ‘program’.

M planning on running a full optimize when I get the chance, but I’m also wondering what else could be going on which would cause this problem. Might need to consult ‘top’!