http://sagenb.org

[William Stein]

Hi,

I've done a rather drastic change to http://sagenb.org, which could
reduce in loss of data to users during the last 16 hours. However,
right now http://sagenb.org should be faster and more robust than the
sorry state it was in before (feedback welcome).

Don't worry, the lost data mentioned above is all still there -- I've
temporarily made http://www.sagenb.com/ (not the .com, which I'm
hopeful nobody ever used) point to the old version of sagenb.org,
which hasn't had any data loss.

William

--
William Stein
Associate Professor of Mathematics
University of Washington
http://wstein.org

[William Stein]

Hi,

Now that sagenb.org can suddenly handle more of a load, it's
interesting to watch the usage.

This command gives the number of *active* users (since processes
timeout very quickly):

os.system('ps -u sagenbws|grep python|wc -l')

There are 31 active simultaneous users right now as I write this.

When I started up the new server, 90 people had logged in within a
matter of minutes. There are now just shy of 20000 accounts on
http://sagenb.org.

-- William

[kcrisman]

Just out of curiosity, what was the change - one which will propagate
to all servers eventually (if they upgrade), making it even easier to
serve courses with them? Just wondering.

- kcrisman

On Nov 24, 10:51 am, William Stein <wst...@gmail.com> wrote:
> Hi,
>

> I've done a rather drastic change tohttp://sagenb.org, which could

> reduce in loss of data to users during the last 16 hours.  However,

> right nowhttp://sagenb.orgshould be faster and more robust than the

> sorry state it was in before (feedback welcome).
>
> Don't worry, the lost data mentioned above is all still there -- I've

> temporarily madehttp://www.sagenb.com/(not the .com, which I'm

[William Stein]

Hi,

I've made some changes to http://sagenb.org as I mentioned before, and
it is indeed for faster. Stress test it! In case you stopped using
http://sagenb.org because it seemed to slow, give it a shot. I'm
curious what it takes before it slows down substantially.

Harald and I are logging numbers of user accounts, number of active
users, etc. This morning about 200 people have used the notebook,
and a snapshot of our log looks like this:

19476 27 1259090090.77 2009-47 Tue_Nov_24_11:14:50_2009
19476 28 1259090120.82 2009-47 Tue_Nov_24_11:15:20_2009
19476 30 1259090150.86 2009-47 Tue_Nov_24_11:15:50_2009
19477 30 1259090180.91 2009-47 Tue_Nov_24_11:16:20_2009
19477 31 1259090210.96 2009-47 Tue_Nov_24_11:16:50_2009

19477 = number of registered users, 31 = active worksheet processes
(the timeout after 2 minutes of inactivity), and the rest is timestamp
info.

- William