Excessive expireover time?
Katsuhiro Kondou
kondou at nec.co.jp
Thu Jul 20 22:48:58 UTC 2000
In article <yl4s5k4jlz.fsf at windlord.stanford.edu>,
Russ Allbery <rra at stanford.edu> wrote;
} > The spool is cnfs, and I don't have "groupbasedexpire" in the inn.conf
} > file (docs say groupbasedexpire defaults to true).
}
} Hm. In that case, I'm inclined to suspect either swapping or something
} odd going on with your buffindexed buffers, but I'm not really qualified
} to debug the latter (having not used buffindexed myself yet).
This may happen.
Here is the result of mine which is one of our upcoming new servers
in production. This is also cnfs and 'groupbasedexpire: false'.
Expire messages:
expire begin Fri Jul 21 02:36:30 JST 2000: (-v1 -z/var/news/log/expire.rm)
Article lines processed 3350210
Articles retained 2692840
Entries expired 657370
Articles dropped 0
Old entries dropped 683785
Old entries retained 72
expire end Fri Jul 21 02:39:20 JST 2000
all done Fri Jul 21 02:39:20 JST 2000
expireover start Fri Jul 21 02:39:20 JST 2000
expireover end Fri Jul 21 04:20:34 JST 2000
lowmarkrenumber begin Fri Jul 21 04:20:34 JST 2000: (/var/news/log/expire.lowmark)
lowmarkrenumber end Fri Jul 21 04:20:38 JST 2000
# inndf -o
3 % overview space used
# inndf -n
3248936 overview data stored
Result of sar;
02:20:00 10 9 28 53
02:40:00 40 9 20 31
03:00:01 11 10 79 0
03:20:00 16 14 71 0
03:40:00 11 12 77 0
04:00:00 11 12 77 0
04:20:01 10 11 79 0
04:40:00 8 7 30 55
The box is Solaris2.6 on Netra t1405(1CPU, 1GBmem, 180GB(RAID-7)).
--
Katsuhiro Kondou
More information about the inn-workers
mailing list