experiences with 2.5.40 on a busy usenet news server

From: Miquel van Smoorenburg (miquels@cistron.nl)
Date: Tue Oct 08 2002 - 03:46:20 EST

Just FYI:

So I booted 2.5.40 with the raid0 fix on our usenet news peering
server yesterday. It is a box that exchanges binary feeds with
about 40 peers, 400 GB/day in, 600 GB/day out.

It's a dual PIII/450, 1 GB RAM, 4x18 GB article spool directly
on partitions (not raw, but normal partitions). INN-2.4/CNFS.

With 2.4.19, it runs fine. With 2.5.40, it goes wildly into
swap. I'm assuming the I/O is pushing the newsserver binaries
and database mappings into swap.

# free
             total used free shared buffers cached
Mem: 1033308 1027316 5992 0 836884 29776
-/+ buffers/cache: 160656 872652
Swap: 976888 364032 612856

No need to swap 364 MB when there's 872 MB still free...
This makes the machine dogslow. An 'expire' process that
runs every night normally takes 15 minutes to finish now
has been running for 10 hours and its still not finished.

Article acceptance rate has halved, the machine can't keep up
with the binaries it is fed.

I'm going to risk corrupting the databases and reboot back
to 2.4.19 now.


To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

This archive was generated by hypermail 2b29 : Tue Oct 15 2002 - 22:00:23 EST