Khemara Lyn wrote:
> Dear All,
>
> I got a lot of the following logged messages in my cache.log file:
>
> 2010/05/10 19:23:59| httpReadReply: Excess data from "GET
> http://ff.search.yahoo.com/gossip?command=..."
> 2010/05/10 19:24:00| httpReadReply: Excess data from "GET
> http://ff.search.yahoo.com/gossip?command=..."
>
> (Sorry, I snipped of the search string and replace with the 3 dots: ...
> confidentiality)
>
> Is there a way to resolve that?
>
> Or, if not, I'd like to know if we can make Squid behave in pass-through
> mode for certain domains, such as search.yahoo.com? In other words, can
> we config an ACL for certain domains such that: Squid just forwards a
> request from the client straight to the origin server without parsing
> the request and saying if there is "Excess data..." or not?
No "excess data" is a poisoning attack. Letting it past results in
serious security breaches affecting all users of the proxy.
Whether intentional or not it is a corruption of reply data in the HTTP
connection and need to be fixed at the source of the problem.
>
> I also observe these:
>
> 2010/05/10 19:34:36| WARNING: unparseable HTTP header field {GET /
> HTTP/1.1}
> 2010/05/10 19:24:00| clientTryParseRequest: FD 5013 (snipped
> client_ip:port) Invalid Request
>
> Any fixes too?
Now that error, if you look at the trace should be fixable. Simply find
out which header is broken and figure out if the fix needs to be done in
in the client software or Squid (usually its a broken java applet or
dynamic web page).
>
> I am running Squid-2.7.STABLE9 on Fedora 12
> (kernel-2.6.32.11-99.fc12.x86_64).
>
> Appreciate for any help.
>
> Thanks & regards,
> Khem
>
Amos
-- Please be using Current Stable Squid 2.7.STABLE9 or 3.1.3Received on Fri May 14 2010 - 05:21:26 MDT
This archive was generated by hypermail 2.2.0 : Fri May 14 2010 - 12:00:06 MDT