[SQU] any bugs with 2.2.17 & squid ?

From: Murat CANDAN <mhcandan@dont-contact.us>
Date: Tue, 03 Oct 2000 21:33:36 +0300

Hi,

        I have been using 2.2.17 on a dual p2-550 box with 8 + 8 GB of scsii as
cache disks for a while.
The problem that i frequently face with squid is that it runs out of
file descriptors (I think).
I have modified /usr/include/bits/types.h
as
#define __FD_SETSIZE 32768
and did ulimit -HSn 32768 with

echo 1024 32768 > /proc/sys/net/ipv4/ip_local_port_range.
Later I configured and compiled and reinstalled squid
but the problems remains with:

2000/10/03 17:59:37| comm_accept: FD 11: (104) Connection reset by peer
2000/10/03 17:59:37| httpAccept: FD 11: accept failure: (104) Connection
reset by peer

As i wrote, there two 8 gigs of disks. When I use both two as disk for
caching
the preiod of squid's execution and start of errors is much small.

Any ideas about problem, kernel spesific or such an other sitiuation ?

p.s. I use squid 2.3 stable 3, patched with 3 announced patches.

--
To unsubscribe, see http://www.squid-cache.org/mailing-lists.html
Received on Tue Oct 03 2000 - 09:32:01 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:55:40 MST