[squid-users] Tcp connection failed problem

Hi,

I'm coming back with this prolem I haven't solved yet. I'm managing an
architecture
of about 20 proxies under squid (most of them version 2.5 Stable1 and
RH9) with three
level of peering :
                    level1
                       ||
                    level 2
                      ||
                    level 3
On every child of one level, I get the message : TCP connection to
parent/8080 failed
very frequently (every 10 or 20 minutes during working hours), the
parent beeing a squid
box of the superior level.
I'm monitoring with mrtg several points like cpu usage, http hits and
requests, disk usage,
unused file descriptor, ... on every proxy, but nothing seems to explain
these failures.
I'm sure of several things during these failures:
- the squid parent is up and running (ping ok without any packet loss)
- it's not overloaded (I get these messages even beetween a child and a
2.8 Ghz Xeon parent
having about 40 http req/s) For info, every squid box is at least a
Proliant DL360 1Ghz with
512 MB RAM.
- no warning messages in cache.log of the parent
- the network connectivity (switch/squid box) is OK and set up to 100
Full

I'm quite worried about this problem because I've seen other person
having this problem but
no clue on how to resolve it. What's more, I'd like to load balance our
internet proxies but
with this problem, it's not possible. 'Cause that would say configuring
only one parent
proxy on the children (the virtual address), and as we use never_direct
on the children, during
connection failures that would result in a "unable to relay" message for
the users.

If anybody can help me... It would be a defeat for me having to switch
to ISA proxies ;-)
Thanx in advance

        Pierre-E
Received on Fri Mar 04 2005 - 02:24:43 MST