vollkommen@xxxxxxx wrote:
Since v3.0 doesn't have IPv6 support, I set up the Polipo proxy (http://www.pps.jussieu.fr/~jch/software/polipo/) as a 6to4 relay in front of squid, by adding "cache_peer 127.0.0.1 parent 8123 3130" to squid.conf. I was able to browse multiple IPv6 sites such as ipv6.googe.com with 3.0.STABLE14-20090416 as the proxy in Firefox, for quite a while, before squid's cache.log spit out: 2009/04/19 20:10:20| Ready to serve requests. 2009/04/19 20:12:09| WARNING: Probable misconfigured neighbor at 127.0.0.1 2009/04/19 20:12:09| WARNING: 143 of the last 150 ICP replies are DENIED 2009/04/19 20:12:09| WARNING: No replies will be sent for the next 3600 seconds 2009/04/19 20:12:31| Detected DEAD Parent: 127.0.0.1 2009/04/19 20:12:52| ipcacheParse: No Address records in response to 'www.ipv6.sixxs.net' After building 3.0.STABLE14-20090419, I find the above issue surfaced much faster than with STABLE14-20090416. Since 20090416 is no longer downlable, I built STABLE14 dated 20090411. It lasted about as long as 20090416, but spit out slightly different logs: 2009/04/19 20:57:36| Ready to serve requests. 2009/04/19 20:59:01| ipcacheParse: No Address records in response to 'ipv6.google.com' 2009/04/19 21:00:01| ipcacheParse: No Address records in response to 'ipv6.google.com' 2009/04/19 21:00:30| ipcacheParse: No Address records in response to 'www.ipv6.sixxs.net' 2009/04/19 21:00:39| 95%% of replies from '127.0.0.1' are UDP_DENIED Could this issue be mitigated from the squid end, so it keeps talking to Polipo (which appears to remain functional)?
The 'no records' issue is appearing due to Squid-3.0 trying to go DIRECT to the IPv6-only website. It reports doing so because of an extremely high connection failure rate (95%) going through the peer on port 8123 and port 3130.
The big clue is "WARNING: 143 of the last 150 ICP replies are DENIED" Try setting cache_peer to use ports "8123 0". Amos -- Please be using Current Stable Squid 2.7.STABLE6 or 3.0.STABLE14 Current Beta Squid 3.1.0.7
