Re: [squid-users] Error making squid-3.1.0.7-20090412 on Mac OS X 10.4

vollkommen_at_gmx.net wrote:
> -------- Original-Nachricht --------
>>> 1. DNS lookups fail completely. 204.127.198.4 is among a handful of
>>> external DNS servers I use, whether defined as dns_nameservers in
>>> squid.conf or picked by squid automatically from /etc/resolv.conf:
>>>
>>> 2009/05/20 21:44:49| comm_udp_sendto: FD 8, (family=2) 204.127.198.4:53:
>>> (22) Invalid argument
>>> 2009/05/20 21:44:49| idnsSendQuery: FD 8: sendto: (22) Invalid argument
>>> 2009/05/20 21:46:10| comm_set_tos: setsockopt(IP_TOS) on FD 7: (22)
>>> Invalid argument
>> Is IPv6 dual-stack enabled in your OS? it looks like the v6 sockets are
>> unable to do the v4-mapping they need to handle v4 traffic.
>> I find it a bit weird that setting TOS also fails, its one of the older
>> functionality and shared between both IP protocols.
>>
>
> I'd think so. I'm using 6to4, if that matters. Here's the snippet from config.log:t
>
> configure:25519: checking whether to enable IPv6
> configure:25534: result: yes
> configure:25548: checking if PF_INET6 is available
> configure:25583: g++ -o conftest -g -O2 -g conftest.cpp >&5
> configure:25587: $? = 0
> configure:25593: ./conftest
> configure:25597: $? = 0
> configure:25599: result: yes
> configure:25624: result:
> configure:25643: checking for IPv6 split-stack requirement
> configure:25666: result: no
> configure:25672: checking for sin6_len field in struct sockaddr_in6
> configure:25703: g++ -c -g -O2 conftest.cpp >&5
> configure:25710: $? = 0
> configure:25727: result: yes
> ...
> #define USE_IPV6 1
> #define IPV6_SPECIAL_V4MAPPED 1
> #define IPV6_SPECIAL_SPLITSTACK 0
>
> Here's cache.log with debug_options 0,5 78,9:
>
> 2009/05/20 23:36:23.570| idnsALookup: buf is 36 bytes for www.ipv6.sixxs.net, id = 0x7fca
> 2009/05/20 23:36:23.570| comm_udp_sendto: FD 7, (family=2) 18.70.0.160:53: (22) Invalid argument
> 2009/05/20 23:36:23.570| idnsSendQuery: FD 7: sendto: (22) Invalid argument
>
>>> 2. Does that above have anything to do with the failure to apply this
>>> patch?
>>>
>>> patch -p0 < b9691.patch
>>> patching file src/dnsserver.cc
>>> Hunk #1 FAILED at 427.
>>> Hunk #2 FAILED at 436.
>>> 2 out of 2 hunks FAILED -- saving rejects to file src/dnsserver.cc.rej
>> I hope not. That patch does not alter anything functionally and is
>> internal to the obsolete dnsserver helper.
>>
>>> 3. I see the following in cache.log. I realize they're warnings, but is
>>> there any way to make squid "happy"? I thought the slash format is a
>> CIDR
>>> mask.
>>>
>>> WARNING: Netmasks are deprecated. Please use CIDR masks instead.
>>> WARNING: IPv4 netmasks are particularly nasty when used to compare IPv6
>> to
>>> IPv4 ranges.
>>>
>>> squid.conf:
>>> acl sitelocalsrc src fec0::/48
>>> acl linklocalsrc src fe80::/64
>> The conf is correct, if the warnings are appearing on that line something
>> has definitely broken.
>>
>> Tomorrow is my actually scheduled day for squid maintenance. I will have
>> to look at these closer after the maintenance work.
>>
>> Amos
>>
>
> Turns out my previous conjecture was wrong. Those warnings appear even after I've commented out all the IPv6 bits in the ACLs. Here' the complete lines from cache.log, in case it matters:
>
> 2009/05/20 23:46:02| WARNING: Netmasks are deprecated. Please use CIDR masks instead.
> 2009/05/20 23:46:02| WARNING: IPv4 netmasks are particularly nasty when used to compare IPv6 to IPv4 ranges.
> 2009/05/20 23:46:02| WARNING: For now we assume you meant to write /12
> 2009/05/20 23:46:02| WARNING: Netmasks are deprecated. Please use CIDR masks instead.
> 2009/05/20 23:46:02| WARNING: IPv4 netmasks are particularly nasty when used to compare IPv6 to IPv4 ranges.
> 2009/05/20 23:46:02| WARNING: For now we assume you meant to write /12

Sorry I didn't get to this a few days ago.

Are you using the default squid.conf that gets built with 3.1 or an
altered version?
Problem #3 will show up for some settings even though there is not yet
any other way to do them than netmasks. So it needs to be looked at but
in some cases can be ignored.

Problem #1, I'm still thinking about.

Amos

-- 
Please be using
   Current Stable Squid 2.7.STABLE6 or 3.0.STABLE15
   Current Beta Squid 3.1.0.8 or 3.0.STABLE16-RC1