Load problems with 2.3 server.

[Simon Lyall]

I'm a little stuck on this one, I'm seeing load problems at a level of
readers well below what I would expect.

Server is 833Mhz Pentium, 1GB RAM with a root drive and /usr/local/news
being a 500GB single external RAID (connection speed 80.0 Mbyte/sec
according to the machine) which doesn't appear to be totally maxed out
yet. Running 2.2.16 Linux with ReiserFS version 3.5.22 on the News
partition. There are 220 odd cycbuff files (damn 2Gig limit).

Running stock 2.3 with buffindexed and cnfs.  Currently it's only getting
a fairly complete non-binaries feed plus an incomplete non-binaries feed
of around 15-20GB/day. History file is 480MB. storeonxref is false but I
have put in "ovgrouppat: *,!control.cancel" in the last couple of days.
expire is still taking 6 hours however.

Problem appears to be that whenever I get above 40 customers the machine
starts going into swap and useability goes right down. Load sneaks
up to about 6 and Readers take about 1 minute to connect and the thing
basicly goes sluggish. If I bounce the server it usually comes right for a
while but eventually goes bad again. I'm a bit worried cause I really need
to handle a fullish feed and a couple of hundred readers which I thought
should be okay. The innd process is very large most of the time ( 998MB
right now with 816MB resident).

The machine also consumes open files like there is no tomorrow, I have
bunmped the max of these up to 32768 which appears to be enough for now,
right now it's using 12,000 odd with only 30 readers and 20 feed
connections.

# cat /proc/sys/fs/file-nr  
18530   12229   32768

Does anybody have any pointers to possible solutions (hopefully other than
adding more RAM)? .

-- 
Simon Lyall.                |  Newsmaster  | Work: simon.lyall at ihug.co.nz
Senior Network/System Admin |              | Home: simon at darkmere.gen.nz
ihug, Auckland, NZ          | Asst Doorman | Web: http://www.darkmere.gen.nz