Re: [squid-users] caching failed tcp connects to destination ips

From: Amos Jeffries <squid3_at_treenet.co.nz>
Date: Thu, 25 Jul 2013 19:17:32 +1200

On 25/07/2013 6:52 p.m., Dieter Bloms wrote:
> Hi,
>
> we use ipv4 and ipv6 tcp protocol for our outgoing interface.
> The most sides are accessable via ipv6, if a AAAA Record is available,
> so ipv6 works great in most cases.
>
> Some sides like http://www.hsp-steuer.de/ announce ipv6 records, but are
> not accessable via ipv6.

Send them a bug report?

> Is it possible that squid notice this fail so that future request will
> go to ipv4 directly and the user doesn't have to wait for the long
> tcp timeout every time ?

Yes it is possible and Squid already does.
If you check your cachemgr "ipcache" report you can see this as the DNS
results domain/IP mapping list OK/BAD flags on each IP address known.
BAD will not be used, OK will be tried, success is always a gamble.

> Maybe with a timestamp, so that it will be refreshed after x hours.

The DNS lookup result TTL is used, whereupon the DNS server is expected
to give better working results. Or if all possible IP (both types) are
tried and all fail the markers are reset and it may be re-tried by some
other request.

Amos
Received on Thu Jul 25 2013 - 07:17:37 MDT

This archive was generated by hypermail 2.2.0 : Thu Jul 25 2013 - 12:00:10 MDT