:: 02-Oct-2000 03:08 (Monday) ::

As I’m sure many of you are aware, there was a problem with all the
statsruns last night (someone must have told the box that I’d be out of
town). I’m working on it now, and the OGR stats will hopefully be up in
an hour or so. I’ll run the RC5 stats immediately after that, so everything
should be fine in the morning.

:: 20-Sep-2000 21:41 (Wednesday) ::

Something got botched when I re-ran OGR stats for 9/18, so I’ll be
re-running them again. }:8/

:: 19-Sep-2000 19:52 (Tuesday) ::

Stats will be late again today :( The automation got a bit cowfused last
night, so I had to manually start the run, and I didn’t notice the problem
until 1000GMT or so. The problem with that is stats take a *very* long
time to run when they are competing with users for disk/CPU, so they take
forever when they run during the day. If they don’t finish soon, I’ll shut
off apache for a while so that they can finish.

Sorry for the inconvenience.

:: 18-Sep-2000 19:01 (Monday) ::

The RC5 statsrun didn’t happen last night due to a problem with the
automated log transfer. I’ll start them running in a few minutes. Sorry
for the inconvenience.

:: 06-Sep-2000 16:52 (Wednesday) ::

The RC5 run is taking quite a long time since it has to do two days, so
I’ve shut off web access to stats for a while to speed things up a bit.
I’ll turn it back on in a little while. Sorry for the inconvenience.

:: 05-Sep-2000 21:49 (Tuesday) ::

Both statsruns failed last night, for different reasons.

I recently removed an index from the RC5 master table with the hope of
saving some space and speeding some things up. As it turns out, that
index is critical for one of the steps in the RC5 statsrun. That’s why
Sept 3 stats ran for almost 24 hours. The index is back in place, and RC5
stats will play ‘catch-up’ tonight as stats for both Sept 4 and 5 will
run.

In OGR, somehow some blocks were assigned to participant 0. Problem is,
there is no participant 0. I’m re-running OGR stats right now; they should
be up very soon.

:: 05-Sep-2000 20:02 (Tuesday) ::

Had some issues with stats last night; I’m working on them now and will
post more info as soon as I can.

:: 01-Sep-2000 12:51 (Friday) ::

Chris Hessmann spotted a problem with team joins in OGR that I wanted to
point out so that people don’t get confused. In a nutshell, the new stats
code isn’t properly handling new team joins.

When a participant joins a team, any old work that was not assigned to a
team is supposed to go to the team they just joined. This is happening
properly in the master table, but it isn’t being reflected in the other
team tables.

Unfortunately, there isn’t a fast easy way to fix this, so we’ll have to
live with it for now. When this is fixed however, all the work that isn’t
currently showing up on the teams will magically appear, and things will
all be as they should, thanks to the fact that the proper info *is* being
stored in Email_Contrib.

The bug report for this is at
http://n0cgi.distributed.net/bugs/show_bug.cgi?id=1655

Sorry for the confusion.

:: 29-Aug-2000 12:20 (Tuesday) ::

The RC5 statsrun died for some reason last night, probably due to a database
deadlock. I’m re-running them by hand right now.

:: 22-Aug-2000 08:54 (Tuesday) ::

I’m working on an improvement to our current team join code. The new code
will speed up the statsrun, and will make things a bit more accurate (the
current system bases team membership on a participant’s team setting at
the time of the statsrun, instead of noting the date that someone joins
a team and using that info to determine what team someone belonged to on
a given date).

Unfortunately, in order to do this, I have to disable team joins until
the conversion is complete.

As soon as the new join process is in place, I’ll be able to re-do the
statsrun to make use of the new data. Team joins will be re-enabled before
I tackle the statsrun though.