On Mon, Apr 25, 2005 at 09:54:12PM -0400, Theo Van Dinter wrote:
> I've done 1 run at 14 days with the patch, and no problems found.  I'll run
> another time or two and see what happens.  It takes a good couple of hours to
> do a single 14 day run though. <sigh>

The second run had no errors either.  I have not verified that I get the
same (or near same) results though.  I'm rerunning without the patch as
well, just to see if it was a random occurance before.

The patch does make things *much* slower though, around 3x:

With (~3h):
status: starting run stage                               now: 2005-04-25 
08:55:58 PM
status:  10% ham: 337    spam: 2232   date: 2005-04-13   now: 2005-04-25 
09:15:27 PM
status:  20% ham: 674    spam: 4464   date: 2005-04-14   now: 2005-04-25 
09:34:16 PM
status:  30% ham: 1011   spam: 6696   date: 2005-04-15   now: 2005-04-25 
09:53:43 PM
status:  40% ham: 1347   spam: 8929   date: 2005-04-17   now: 2005-04-25 
10:14:25 PM
status:  50% ham: 1684   spam: 11161  date: 2005-04-18   now: 2005-04-25 
10:33:53 PM
status:  60% ham: 2021   spam: 13393  date: 2005-04-20   now: 2005-04-25 
10:53:15 PM
status:  70% ham: 2358   spam: 15625  date: 2005-04-21   now: 2005-04-25 
11:12:21 PM
status:  80% ham: 2695   spam: 17857  date: 2005-04-22   now: 2005-04-25 
11:31:29 PM
status:  90% ham: 3031   spam: 20090  date: 2005-04-23   now: 2005-04-25 
11:50:28 PM
status: 100% ham: 3367   spam: 22314  date: 2005-04-25   now: 2005-04-26 
12:09:24 AM
status: completed run stage                              now: 2005-04-26 
12:09:29 AM

Without (~1h):
status: starting run stage                               now: 2005-04-25 
04:28:05 PM
status:  10% ham: 338    spam: 2231   date: 2005-04-12   now: 2005-04-25 
04:34:47 PM
status:  20% ham: 675    spam: 4463   date: 2005-04-14   now: 2005-04-25 
04:40:59 PM
status:  30% ham: 1010   spam: 6697   date: 2005-04-14   now: 2005-04-25 
04:47:13 PM
status:  40% ham: 1348   spam: 8928   date: 2005-04-17   now: 2005-04-25 
04:53:57 PM
status:  50% ham: 1685   spam: 11160  date: 2005-04-18   now: 2005-04-25 
04:59:41 PM
status:  60% ham: 2022   spam: 13392  date: 2005-04-20   now: 2005-04-25 
05:05:41 PM
status:  70% ham: 2358   spam: 15625  date: 2005-04-21   now: 2005-04-25 
05:11:14 PM
status:  80% ham: 2695   spam: 17857  date: 2005-04-22   now: 2005-04-25 
05:16:55 PM
status:  90% ham: 3032   spam: 20089  date: 2005-04-23   now: 2005-04-25 
05:22:06 PM
status: 100% ham: 3367   spam: 22314  date: 2005-04-25   now: 2005-04-25 
05:27:35 PM
status: completed run stage                              now: 2005-04-25 
05:27:36 PM


Without the patch, lots of issues starting after 80%.

I'm thinking that a per-request ID isn't necessary, but a per instance
ID is -- either initialized at check() or whenever the module gets a
new instance setup.

-- 
Randomly Generated Tagline:
A writer thinks of critics as a tree feels about dogs.

Attachment: pgpVfvVqaitwX.pgp
Description: PGP signature

Reply via email to