Expire performance with big history and spool

[N]

James Brister wrote:

> [ does anyone these days have anything but a big history and spool? ]

Yes, but the problems encountered then are usually of the "I can't connect
/ post binary attachments, what's wrong?" variety.


> What sort of times are people seeing for expire to run on a reader box
> running Solaris with 800+ Mb of history and 150Gb of CNFS spread
> across raw disks partitions?

Well, FreeBSD with 200 GB; ~news/db/history is 1.6 GB, from last night's
news.daily run:

expire begin Sat Oct 30 02:39:15 CEST 1999: (-v1 -z/usr/local/news/log/expire.rm )
    Article lines processed 17678866
    Articles retained       17024528
    Entries expired           654338
    Articles dropped               0
    Old entries dropped       863825
    Old entries retained      584469
expire end Sat Oct 30 03:01:43 CEST 1999
        all done Sat Oct 30 03:01:43 CEST 1999
        expireover start Sat Oct 30 03:01:44 CEST 1999
        expireover end Sat Oct 30 04:56:52 CEST 1999

On a machine with a bit less memory and about half the spool but with INN
2.2 (instead of 2.3), expire takes almost three hours to run.


> Does anyone have a bigger setup than this? (or is that a really naive
> question)?

Yes, unfortunately...


	-- Niels.