[squid-users] How to catch a problem

 
I have problem from 2 days...
A lot of messages:

> comm_accept: FD 24: (53) Software caused connection abort
>2005/04/06 15:21:48| httpAccept: FD 24: accept failure: (53) Software
caused connection abort

after while squid restarts with msg

>FATAL: xcalloc: Unable to allocate 1 blocks of 4104 bytes!

>Squid Cache (Version 2.5.STABLE9): Terminated abnormally.

How to catch who is produce this error... In the post bellow Henrik explain
that is Reset problem... But I starting TCDump find more R packets.....
From too many sources from my net...

>>> tcpdump dst proxy and port 8080 | grep R
15:26:24.651725 xxxxx.1864 > proxy.8080: R 326:326(0) ack 1868 win 0 (DF)
15:26:25.866586 xxxxx.2139 > proxy.8080: R 8527:8527(0) ack 8343 win 0 (DF)
15:26:25.866661 yyyy.2143 > proxy.8080: R 1572:1572(0) ack 1546 win 0 (DF)
15:26:29.485790 zzzzz.1727 > proxy.8080: R 19479612:19479612(0) win 0 (DF)

Is there a way to locate more acurate who make this problem... And how to
isolate???
I have a big troubles with this... Too often proxy restarting:(((((

TIA

P.S. Here is the post from arhive....

>From: Henrik Nordstrom <hno@dont-contact.us>
>Date: Fri, 15 Feb 2002 12:13:45 +0100

>This message seems to indicate you have clients who aborted the
>connection before it was even fully established, during the TCP
>handshake, and your OS now signals such connection attempts to the
>application with an error.

>A tcpdump should tell you pretty quickly if this is indeed the case.
>Simply run 'tcpdump "your.proxy.ip.address" "proxy_port"' for a short
>while until you see one of these messages, and then look for RST
>packets in the dump.
Received on Wed Apr 06 2005 - 06:31:54 MDT