Slow expireover (again)

Mike Zanker mike-sender-6677e0 at zanker.org
Sun May 23 08:30:19 UTC 2004


On 23 May 2004 01:09 -0700 Russ Allbery <rra at stanford.edu> wrote:

> Did they go down in direct proportion to the number of articles on the
> system, or did they start going downhill later on after the article
> count had stabilized?

More or less in direct proportion to article lines processed:

Articles	expireover
--------	----------
273320		27m
824054		1h 50m
1026416		1h 47m
1239219		2h 10m

(added about 20,000 groups to active)

1430128		2h 33m
1634668		2h 55m
1884172		3h 42m
2228048		4h 16m

(machine rebooted)

2525160		6h 57m


I have just found a news.daily from February when this box was using 
timecaf but had approximately 20,000 groups in active. There were 
2612176 article lines processed and expireover took only 22 minutes!

I have now changed from timehash to timecaf to see if the difference in 
expireover behaviour affects the times.

Mike.


More information about the inn-workers mailing list