Hello ludovit,
We experienced the similar problems of crashing squid with "signal 6 and status 0" symptoms in the /var/log/messages. There were hundreds of crashes per hour during the working hours. There were also many crashes during night hours. And I had no idea what the cause of the problem was.
But your case shed light on the problem.
I captured network traffic and saw in the dump many requests to IPv6 addresses like this: POST http://[2001:67c:4e8:f004::a]:80/api HTTP/1.1\r\n
I think that the requests were generated by the Telegram application, because neighboring similar requests were addressed to Telegram IPv4 pools: POST http://91.105.192.100:80/api HTTP/1.1\r\n
To fix the problem somehow, I added the following rules to the very beginning of the SQUID configuration:
acl urldst_ipv6 url_regex ^http://\[
http_access deny urldst_ipv6
Now it seems the problem has been solved.
I was wondering if this is a SQUID bug. In my opinion, SQUID should not crash on requests to IPv6 resources, but should send a response with an error, for example, Bad gateway 502, or Not Acceptable 406.
Kind regards,
Ankor.
We experienced the similar problems of crashing squid with "signal 6 and status 0" symptoms in the /var/log/messages. There were hundreds of crashes per hour during the working hours. There were also many crashes during night hours. And I had no idea what the cause of the problem was.
But your case shed light on the problem.
I captured network traffic and saw in the dump many requests to IPv6 addresses like this: POST http://[2001:67c:4e8:f004::a]:80/api HTTP/1.1\r\n
I think that the requests were generated by the Telegram application, because neighboring similar requests were addressed to Telegram IPv4 pools: POST http://91.105.192.100:80/api HTTP/1.1\r\n
To fix the problem somehow, I added the following rules to the very beginning of the SQUID configuration:
acl urldst_ipv6 url_regex ^http://\[
http_access deny urldst_ipv6
Now it seems the problem has been solved.
I was wondering if this is a SQUID bug. In my opinion, SQUID should not crash on requests to IPv6 resources, but should send a response with an error, for example, Bad gateway 502, or Not Acceptable 406.
Kind regards,
Ankor.
ср, 12 окт. 2022 г. в 15:30, Ludovit Koren <ludovit.koren@xxxxxxxxx>:
>>>>> Amos Jeffries <squid3@xxxxxxxxxxxxx> writes:
> On 12/10/22 21:12, Ludovit Koren wrote:
>> Hi,
>> I am running squid-5.7 on FreeBSD 12.3-STABLE r371879. Occasionally
>> I
>> get the following error:
>>
>> #3 0x000000080111fcb1 in __assert (func=<optimized out>, file=<optimized out>, line=<optimized out>, failedexpr=<optimized out>) at /usr/src/lib/libc/gen/assert.c:51
>> #4 0x0000000000698fcd in Ip::Address::getAddrInfo (this=0x861c04588, dst=<optimized out>, force=0) at Address.cc:663
>> #5 0x000000000068b732 in comm_openex (sock_type=sock_type@entry=1, proto=proto@entry=6, addr=..., flags=1, note=0x85ce42dc0 "[fe80::21f:29ff:fe28:7017]") at comm.cc:347
> ...
>> The squid is compiled without IPv6 option, so I do not understand
>> why it
>> tries to reach IPv6 address.
>>
> A client is requiring connection to an IPv6 server. But Squid cannot
> convert that IPv6 address or use on an IPv4-only network. Lack of IPv6
> support also forbids IPv6 failover handling being used.
So the solution is to reenable IPv6 in the squid, as well as in the OS
network stack? Am I right?
Thank you.
Regards,
lk
_______________________________________________
squid-users mailing list
squid-users@xxxxxxxxxxxxxxxxxxxxx
http://lists.squid-cache.org/listinfo/squid-users
_______________________________________________ squid-users mailing list squid-users@xxxxxxxxxxxxxxxxxxxxx http://lists.squid-cache.org/listinfo/squid-users
