[squid-users] Squid quit working after swapping routers

Hi All,

I'm running Squid 2.5-stable3 on a FreeBSD system. In addition to the
caching, I'm using it as a half-ass method to off-load bandwidth from our
primary Internet connection.

Our primary circuit was a very slow (128K) fractional T1. Cheap & fast
cable-tv 'net became available here, so I got that in addition to the T1
and installed a 2nd NIC in the Squid server. I configured Squid
with: tcp_outgoing_address 12.1.2.3 to direct its traffic out the
secondary NIC over the cable service. The default route on the server, and
other machines accessing it remained via the T1 gateway router. Everything
worked nicely.

Yesterday, our ISP swapped out the T1 router (old Cisco for a newer Cisco)
and increased our circuit speed. Following that, and a reboot of the Squid
server, Squid no longer works.

Browsers on client PCs hang for a minute or so, then display:
        The following error was encountered:
        Connection Failed
        The system returned:
           (61) Connection refused
        The remote host or network may be down. Please try the request again.

I can't figure out what the heck is broken. There are no errors in the
Squid or system logs. Squid starts and binds its sockets ok. I tried
turning on debugging, but didn't spot anything related to the problem.

If I remove the tcp_outgoing_address directive, it works as a regular
caching server over the T1.

I don't know enough about Ciscos to know what they might have
changed. Neither one had any packet filtering turned on.

TIA for any ideas on this one...

   -Wayne
Received on Fri Jul 30 2004 - 15:48:13 MDT