Core I7 2600K@4Ghz
2x128Gb Intel 510 SSD'd (databases)
Intention is to make some much needed changes to some of the underlying tables and therefore the scripts that parse and create the data. Main reason for this the way I create the stats.
I have two databases, each on a different SSD. First database is where I parse the xml files into and do all the manipulation to. The second database is the web facing one. Don't want to do major inserts/updates
I'll try and post one liners for stuff I add in..
Added a 'last28' field into the boinc users table rather than computing it on the fly. Not quite 3rd normal form for a database but it allows it to be indexed and sorted in the front end. Coded the perl scripts to calculate it and fixed the 'users' php script to allow sorting.
Amended the nightly script which calculates a users 'main' team across boinc projects. This is a bit of a grey area really. The script basically uses
I finally got round to looking at why the numbers for the last 28 days of points/teams/users/hosts on the proj page were crap.
Well, it turns out that ever since I converted my tables to be 'generic' I hadn't altered my nightly rollover script to handle this. So it was doing a generic update on the projects table..
I just fixed it to run through each project and do the updates singly. We'll see hot it works overnight tonight. Of course it will take a while to correct fully..
finally got a parser going. It's kind of manual right now as this is a one time thing, so each run just gets one teams data (users and team). I've done the top 10. Might just write a quick wrapper and pull out the top 200 team numbers which is all that is listed at
Then I'll have to manually do the ranking and such. Project ranks are going to be wrong as they will just be within the database
How I had missed this I don't know. thanks vaughan for pointing it out..
I added in the coding for the indexes on 'yesterday' and 'last_update'. Done the frontend changes for last_update, have yet to do the ones for yesterday.