forwarding loops

Hi guys,

We've recently started upgrading some of our internal caches to
Squid 2.0.RELEASE.

We're seeing lots of error messages like these:

1998/10/06 11:26:10| WARNING: Forwarding loop detected for
'http://www.independent.ie/1998/278/b14e.shtml'
1998/10/06 11:26:10| --> 1.0 squid.eei.ericsson.se:3128 (Squid/2.0.RELEASE), 1.0
stp-proxy.ericsson.se:3128 (Squid/2.0.RELEASE)

- we're all seeing them on our respective caches (spread around Ericsson)

Forwarding loops being detected before were usually caused by something
like two siblings having the same visible_hostname. We don't have that...

We also have the delightful situation:

1998/10/06 11:17:25| --> 1.0 squid.eei.ericsson.se:3128 (Squid/2.0.RELEASE), 1.0
stp-proxy.ericsson.se:3128 (Squid/2.0.RELEASE), 1.0 stp-proxy.ericsson.se:3128
(Squid/2.0.RELEASE), 1.0 squid.eei.ericsson.se:3128 (Squid/2.0.RELEASE), 1.0
stp-proxy.ericsson.se:3128 (Squid/2.0.RELEASE), 1.0 squid.eei.ericsson.se:3128
(Squid/2.0.RELEASE), 1.0 stp-proxy.ericsson.se:3128 (Squid/2.0.RELEASE), 1.0
squid.eei.ericsson.se:3128 (Squid/2.0.RELEASE), 1.0 stp-proxy.ericsson.se:3128
(Squid/2.0.RELEASE), 1.0 squid.eei.ericsson.se:3128 (Squid/2.0.RELEASE), 1.0
stp-proxy.ericsson.se:3128 (Squid/2.0.RELEASE), 1.0 squid.eei.ericsson.se:3128
(Squid/2.0.RELEASE), 1.0 stp-proxy.ericsson.se:3128 (Squid/2.0.RELEASE), 1.0
squid.eei.ericsson.se:3128 (Squid/2.0.RELEASE), 1.0 stp-proxy.ericsson.se:3128
(Squid/2.0.RELEASE), 1.0 squid.eei.ericsson.se:3128 (Squid/2.0.RELEASE), 1.0
stp-proxy.ericsson.se:3128 (Squid/2.0.RELEASE), 1.0 squid.eei.ericsson.se:3128
(Squid/2.0.RELEASE), 1.0 stp-proxy.ericsson.se:3128 (Squid/2.0.RELEASE), 1.0
squid.eei.ericsson.se:3128 (Squid/2.0.RELEASE), 1.0 stp-proxy.ericsson.se:3128
(Squid/2.0.RELEASE), 1.0 squid.eei.ericsson.se:3128

All one line - and it kept going and didn't stop... a bug ?

I also saw a line with three caches going around in a circle. Two
were in Ireland and one in Denmark:

1->2->3->1->2->3->1->2->3...

and 3 didn't even have 1 configured as a sibling !?

We can't see anything blatantly incorrect in our configs - anyone
got any ideas what it might be ?

I've currently set the problematic peers (there's more than one causing
this) to "no-query" and it seems to be a little better. At least we're
not getting the *huge* errors - but the other "normal" forwarding loop
messages are still appearing.

We each have miss_access allowed for our peers - so our users don't
get those annoying access_denied error messages (can't wait until
the only-if-cached stuff is implemented !).

I'm guessing that removing miss_access will fix this immediately
but we don't want to do that !

*ANY* help will be muchly appreciated :-)

Cheers,
Niall

-- 
Niall Doherty          | mailto:ndoherty@eei.ericsson.se
Systems Engineer       | http://www.ericsson.ie
Voice: +353 1 207 7506 | Ericsson Systems Expertise Ltd.,
Fax:   +353 1 207 7115 | Beech Hill, Clonskeagh, Dublin 4, Ireland.