2013-07-02 [tenhou] identity
Well, re: the last entry, it's unfortunately come to this and I've had to drop the entire database. I will now begin migrating arcturus.su to a new server I have purchased, and rebuild the database there. I can't say for sure how long this will take but I will try to get it done reasonably soon because the ranking.pl tool is quite popular.
The new service will offer the same features as the old, along with:
- tracking of actual rank (and some nice graphing I hope)
- possible pseudo-rank/rate for private lobbies, e.g. it will offer a dan and R for L7447
- more integrity (data wise - expect 100% accuracy this time)
- possibly more up-to-date insertion of data, rather than every 24hours
Stay tuned I guess. Also I want to get rid of this hatena blog thing, it's really shitty. I might write something of my own.
Hosting issues mean I have to cut the size of the backend MySQL database pretty much immediately. Right now it stands at 10GB, which has recently been a problem anyway (a source of query timeouts re: the last post, for example). My solution in the interim, in order to maintain at least some functionality, is to remove all entries pre-2012-01-01.
mysql> DELETE FROM `gamerecords` WHERE date < '2012-01-01';
This was a hard decision to make. I don't like the fact that I have to do this so soon - in fact, exams make it quite inconvenient for me to be messing with this stuff - but this is the only reliable short term solution. I have backed up the database already, as of 2012-01-15 (pre-insert, so I guess more like 2012-01-14...).
The database will be back, though. My intention is to move to a new server I have set up, but the infrastructure behind arcturus.su is actually quite complex and not that easy to move without a lot of work. There is also just reams of data, which is quite a serious task to host (I believe the entire site exceeds 60GB in size now, given all the Tenhou data in various forms).
It will happen soon, though. For now we have to put up with limited usefulness of ranking.pl, but my hand is being forced in this regard and there is not much that can be done.
I was going to rebuild ranking.pl's backend database anyway, since it had a few problems, so I guess this is a blessing in disguise because it will force me to do it sooner.
Stay tuned for more updates.
I realised today that when the day's records are inserted into the MySQL database, ranking.pl starts to throw 504 Gateway Timeouts. Considering this process takes over two hours to complete, I will look into correcting this. Kinda busy right now but I have put a notice on the ranking.pl input page so that people are aware that shit isn't totally fucked.
More updates coming soon.
The perl executable path changed its location on my server about two weeks ago, so for about 10 days there was limited service for ranking.pl. It's fixed now and the missed days have been entered into the database.
As a result of this I've made a few changes that make it easier to patch any potential gaps or errors in the database. Since it's really big, it can be the case that entries get entered multiple times - checking for duplicate entries is not possible since games are not necessarily unique. I'm in the process of fixing some gaps or double entries that have plagued the database for quite a while.
Also, the 60 millionth game was recently played. There's a record I guess.
I have the intention (but not necessarily the will) to add support for jansou-mode gametype entry into the database. When I get around to it, I will add all games up to that point, meaning the database will be complete. I harvest the data but I haven't automated its entry into the database yet. I think some database modifications are necessary, which might take a million years.
In related news MySQL is a pile of shit, but a necessary evil.