Justin Popa wrote:
Afternoon everyone, I have a small problem.
I've got a user who needs to access a website, and when he goes there
he occasionally gets an Access Denied error. Looking in the logs, I
see the following:
10.150.6.53 - hoffmand [04/Feb/2008:13:53:33 -0500] GET
http://buymtdonline.arinet.com/EW54MTD/MTDC/Include/cfgCustom.js
HTTP/1.0 200 13276 TCP_MISS:DIRECT
10.150.6.53 - (hoffmand) - [04/Feb/2008:13:53:33 -0500] GET
http://buymtdonline.arinet.com/scripts/EmpartISAPI.dll? HTTP/1.0 403
1403 TCP_DENIED:NONE
10.150.6.53 - hoffmand [04/Feb/2008:13:53:33 -0500] GET
http://buymtdonline.arinet.com/scripts/EmpartISAPI.dll? HTTP/1.0 200
4908 TCP_MISS:DIRECT
Note: In the second line I added the (hoffmand) because it's obviously
his traffic, just not marked as such.
Which indicates Squid did not receive authentication details for that
request.
Now, for the fun stuff. We use
AD for our authentication source and that works great. I've also
looked through our deny statements in squid.conf, of which there are
only 3 and here they are:
1) Blocking based on url. The blocked entries are all like
myspace.com, facebook.com, 2girls1cup.com, etc...
2) Blocking based on streaming media. These entries are like .avi,
.mov, .wmv, etc.
3) Blocking if Active Directory authentication failed.
Any thoughts on what this might be just looking at it? Obviously I'm
sure you guys need more, but any help you can give me in starting to
track down the why would be awesome. Thanks
Squid did not receive authentication details with the first request for
EmpartISAPI.dll, threw the 403 and then (likely*) got the same request
with authentication details. I would assume all this happened with out
the client seeing anything. At least in this instance. I don't know
enough about NTLM authentication to say why the browser would not send
authentication details with that request.
Chris
* With the default squid.conf setting strip_query_terms on there is no
way to tell if that is indeed the same request, but assuming the time
stamps are accurate, it's likely.
