staff blogs

distributed.net staff keep (relatively) up-to-date logs of their activities in .plan files. These were traditionally available via finger, but we've put them on the web for easier consumption.

2010-09-21

Integer boundaries

Filed under: stats,Uncategorized — Tags: , , , @ 15:31 +00:00

So, stats hit a milestone of sorts. We’re got a team that has now done 2^63 keys in rc5. This has overflowed the column that it was stored in.

The temporary solution is to increase these columns to NUMERIC(25), which is 10^25. This gives us over a million times more storage until we get a more “correct” solution, which is to store stats units in the database. But that requires more work, and we’re looking to get things up as quickly as possible.

We apologize for the downtime but we didn’t realize how good you guys are!

2007-11-26

nerf [26-Nov-2007 @ 18:46]

Filed under: project status,stats @ 18:46 +00:00

:: 26-Nov-2007 18:46 GMT (Monday) ::

At our recent code-a-thon in Austin we made some significant headway on a
couple of stats projects. One is more behind the scenes and probably not as
interesting to the public (logdb), while the other is something we’re hoping is
going to be quite useful for a number of you.

In short, we have made a few changes to the way stats system works to enable it
to read either our internal log files or (drumroll) logs from a personal proxy.

We should warn you that this code path is relatively untested, but that’s what
we’re hoping you will help us with. We’re also sadly behind on documentation,
but we’re working on that, too.

We would, as always, be glad to hear about bugs suggestions or improvements
(even better when they come with a diff!). http://bugs.distributed.net/ is the
best place to report these issues. If you have questions, you can either submit
them as a bug or join us on #distributed via IRC on irc.distributed.net.

Link for the tarball:
http://http.distributed.net/pub/dcti/source/dnet-stats-1.0.1.tar.gz
Link for the sig for the tarball:
http://http.distributed.net/pub/dcti/source/dnet-stats-1.0.1.tar.gz.asc

-Nerf

2006-02-15

nerf [15-Feb-2006 @ 19:09]

Filed under: clients @ 19:09 +00:00

:: 15-Feb-2006 19:09 GMT (Wednesday) ::

The speeds calculator has been updated to reflect more modern processors.
Note that this is for rough estimates only.

Further note that I have nothing to do with this code other than being
the one to have it in their home directory. All credit for this latest
version goes to Jonathan Smith from the rc5 mail list.

http://www.distributed.net/~nerf/rc5calc.html

2004-04-21

nerf [21-Apr-2004 @ 15:27]

Filed under: Uncategorized @ 15:27 +00:00

:: 21-Apr-2004 15:27 GMT (Wednesday) ::

A word or two on the status of ogr-verification. As you know, We’re
having to restore from backups as of February. Unfortunately, when that
backup was taken, there was a problem and the ogr-ver data was not up to
date. The end result is that we have several months of logs to go
though to catch up.

I haven’t been able to do many test runs without something else running
on the box, but it looks like it’s going to be quite a bit faster than
the old box. It’s taking around 2-2.5 hours to process a day now, which
is already an improvement and stats are still catching up, so we’re
competing for I/O.

2003-09-17

nerf [17-Sep-2003 @ 11:15]

Filed under: Uncategorized @ 11:15 +00:00

:: 17-Sep-2003 11:15 GMT (Wednesday) ::

Just a note to let everyone know I’m alive, too. I’ve been working on
getting ogr-verify to run in incremental mode. That is, only touching
what has changed since the previous day. While this makes the code a
bit more complex, it should save us several hours of processing per
day.

As a test, I’ll be running the scripts against the logs from the
beginning of OGR, which should give us some interesting graphs.

As always, if there’s something you would like to see, drop me an
email or, better yet, send a patch.

2003-08-16

nerf [16-Aug-2003 @ 09:17]

Filed under: Uncategorized @ 09:17 +00:00

:: 16-Aug-2003 09:17 GMT (Saturday) ::

OGR-24 is almost over. Rounding error shows 100% on the stats page, but
there’s still a few dozen stubs to be done. After some final checking
we will let you know when it’s all done.

2003-07-15

nerf [15-Jul-2003 @ 10:24]

Filed under: Uncategorized @ 10:24 +00:00

:: 15-Jul-2003 10:24 GMT (Tuesday) ::

ogr-verify failed to run last night (permission problem in the database) so
the completion numbers didn’t update. I restarted it and the numbers will
update when it’s done (should be around 18:00 GMT). Sorry about that.

2003-06-10

nerf [10-Jun-2003 @ 15:17]

Filed under: Uncategorized @ 15:17 +00:00

:: 10-Jun-2003 15:17 GMT (Tuesday) ::

Just a few words about how the completion numbers are generated. The
single “Percent Complete” number is actually the average of how far along
we are for pass 1 and pass 2. The current breakdown of how things are
going is:

project_id | pass1_pct | pass2_pct | tot_pct
————+———–+———–+———
24 | 85.23 | 49.70 | 67.46
25 | 81.90 | 47.60 | 64.75

Also, these numbers represent the number of stubs done. Given that some
stubs require much more work than others, the percent complete may not be
a linear representation of how much work is left to do. On the other hand,
the work returned is spread out fairly evenly across the stubspace, so it
should be close.

2003-05-27

nerf [27-May-2003 @ 12:53]

Filed under: Uncategorized @ 12:53 +00:00

:: 27-May-2003 12:53 GMT (Tuesday) ::

Your years of waiting are over. We now have completion data for OGR.

See http://stats.distributed.net/projects.php?project_id=24 (and 25) for
all the fun.

Note that we’re not done. Expect to see a bit more data coming in the
future.

2003-04-14

nerf [14-Apr-2003 @ 21:59]

Filed under: Uncategorized @ 21:59 +00:00

:: 14-Apr-2003 21:59 GMT (Monday) ::

After a bit of help, I got that final query working. Thanks again to
CalicoJak.

You’re probably all now wondering “when will we see the numbers?” Fair
enough question, but I’m afraid the answer is “I don’t know.”

I don’t really know enough PHP to get the connection to the database
working nor have I had the time to sit down and figure it out. Yes, I
know it can’t be that hard.

Older Posts »