Re: Strange squid behavior

From: Duane Wessels <wessels>
Date: Thu, 17 Oct 96 12:08:58 -0700

sakr@Chg.RU writes:

>Dear Squiders,
>
>Time to time my squid send to WWW client the following message:
>
>The following error was encountered:
> Unable to Retrieve
>
>The system returned:
>
> No ICP replies received and the host is beyond the firewall.
>The cache administrator may need to double-check the cache configuration.
>
>This means that:
> The requested URL can not currently be retrieved.
>
>
>But at the same time "Reload" button in Netscape produce correct result.
>Squid is configured as our network is beyond firewall.
>
>I studied log files of both Squid caches - this one and his parent.
>First request to parent produces (according parent log)
>UDP_MISS/000/NONE ICP_QUERY,
>but his child treats it as ERR_CANNOT_FETCH/400/NO_DIRECT_FAIL.
>"Reload" produces GET request to parent (TCP_MISS/200/FIRST_PARENT_MISS reply
>which treats as TCP_MISS/200/SINGLE_PARENT 2347 GET).
>
>It's very interesting but after squid killing and restart it works correctly.
>But after some time (a day) situation may be repeated.
>I guess this squid behavior starts after temporary (>1 hour) line break down
>(but I don't sure).
>
>Squid version - 1.0.16, OS - FreeBSD-2.1.0-RELEASE (both Squid caches).
>
>Could somebody help me reach robust work of Squid.

you must have a sibling neighbor as well as a parent?

if you're behind a firewall and have only one neighbor (a parent) you
should turn on 'single_parent_bypass' in the config.

I don't know enough about your configuration, but I wonder if it
would help to comment out these two lines in neighbors.c:

        /* skip any cache where we failed to connect() w/in the last 60s */
        if (squid_curtime - e->last_fail_time < 60)
            continue;

Duane W.
Received on Thu Oct 17 1996 - 12:08:58 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:33:18 MST