On 10/10/2013 2:19 p.m., Luke Pascoe wrote:
> Hi all,
>
> I've been banging my head against an auth issue with Squid3 for some
> time now, I'm hoping someone here will be able to shine a light on it.
>
> I have installed squid 3 (3.1.20) on Debian 7 (using the Debian package)
>
> I've configured it according to the wiki doc:
> http://wiki.squid-cache.org/ConfigExamples/Authenticate/WindowsActiveDirectory
>
> For the most part, this configuration works great. Browsing through a
> browser and most applications works fine using SSO.
Indicating that it is unlikely a proxy issue. But it might still be
proxy related if the particular apps in question are relying on
something unusual in the protocol.
> However, some applications fail to authenticate correctly. Notably,
> windows update and Microsoft Lync. These attempt to auth using the
> logged in user account, fail, then pop up an authentication dialog.
> Putting correct credentials into the dialog doesn't work either.
>
> Every test I've run suggests that each of the components are working
> correctly. Kerberos is configured correctly, squid_kerb_auth works,
> ntlm_auth works as does wbinfo.
>
> TCPDumps of sessions that work (through a browser, IE, Chrome, FF all
> work) and ones that don't (Windows update) appear pretty much the
> same. In both cases the server provides a challenge and the client
> responds (dumps available if you want them)
>
> Here's what I see in cache.log for a successful auth:
>
> 2013/10/10 14:02:11| negotiate_wrapper: Got 'YR
> TlRMTVNTUAABAAAAl4II4gAAAAAAAAAAAAAAAAAAAAAGAbEdAAAADw==' from squid
> (length: 59).
> 2013/10/10 14:02:11| negotiate_wrapper: Decode
> 'TlRMTVNTUAABAAAAl4II4gAAAAAAAAAAAAAAAAAAAAAGAbEdAAAADw==' (decoded
> length: 40).
> 2013/10/10 14:02:11| negotiate_wrapper: received type 1 NTLM token
> 2013/10/10 14:02:11| negotiate_wrapper: Return 'TT
> TlRMTVNTUAACAAAADgAOADgAAAAVgoniTIZLWbbwZ5wAAAAAAAAAAHAAcABGAAAABgEAAAAAAA9QAEEAQwBJAEYASQBDAAIADgBQAEEAQwBJAEYASQBDAAEADABCAEEAQwBPAE8ATgAEABoAcABhAGMAaQBmAGkAYwAuAGwAbwBjAGEAbAADACgAYgBhAGMAbwBvAG4ALgBwAGEAYwBpAGYAaQBjAC4AbABvAGMAYQBsAAAAAAA=
> '
> 2013/10/10 14:02:11| negotiate_wrapper: Got 'KK
> TlRMTVNTUAADAAAAGAAYAIgAAAAYABgAoAAAAA4ADgBYAAAADAAMAGYAAAAWABYAcgAAABAAEAC4AAAAFYKI4gYBsR0AAAAPLtwz2n5LAwh+0a8fdq1dRVAAQQBDAEkARgBJAEMAbgBiAG8AdwBrAGUASABTAE4AWgAtAE4AQgBPAFcASwBFAHv2o2FAquhEAAAAAAAAAAAAAAAAAAAAAMG+Tip01j5dx+x1+2hdSAjkN+OskudoRIttHfifAVCtWaNrbmpBARM='
> from squid (length: 271).
> 2013/10/10 14:02:11| negotiate_wrapper: Decode
> 'TlRMTVNTUAADAAAAGAAYAIgAAAAYABgAoAAAAA4ADgBYAAAADAAMAGYAAAAWABYAcgAAABAAEAC4AAAAFYKI4gYBsR0AAAAPLtwz2n5LAwh+0a8fdq1dRVAAQQBDAEkARgBJAEMAbgBiAG8AdwBrAGUASABTAE4AWgAtAE4AQgBPAFcASwBFAHv2o2FAquhEAAAAAAAAAAAAAAAAAAAAAMG+Tip01j5dx+x1+2hdSAjkN+OskudoRIttHfifAVCtWaNrbmpBARM='
> (decoded length: 200).
> 2013/10/10 14:02:11| negotiate_wrapper: received type 3 NTLM token
> 2013/10/10 14:02:11| negotiate_wrapper: Return 'AF = nbowke'
>
> And here's what I see for a failed auth:
>
> 2013/10/10 14:02:11| negotiate_wrapper: Got 'YR
> TlRMTVNTUAABAAAAl4II4gAAAAAAAAAAAAAAAAAAAAAGAbEdAAAADw==' from squid
> (length: 59).
> 2013/10/10 14:02:11| negotiate_wrapper: Decode
> 'TlRMTVNTUAABAAAAl4II4gAAAAAAAAAAAAAAAAAAAAAGAbEdAAAADw==' (decoded
> length: 40).
> 2013/10/10 14:02:11| negotiate_wrapper: received type 1 NTLM token
> 2013/10/10 14:02:11| negotiate_wrapper: Return 'TT
> TlRMTVNTUAACAAAADgAOADgAAAAVgoniHY15O8+h+dcAAAAAAAAAAHAAcABGAAAABgEAAAAAAA9QAEEAQwBJAEYASQBDAAIADgBQAEEAQwBJAEYASQBDAAEADABCAEEAQwBPAE8ATgAEABoAcABhAGMAaQBmAGkAYwAuAGwAbwBjAGEAbAADACgAYgBhAGMAbwBvAG4ALgBwAGEAYwBpAGYAaQBjAC4AbABvAGMAYQBsAAAAAAA=
> '
> 2013/10/10 14:02:11| negotiate_wrapper: Got 'YR
> TlRMTVNTUAADAAAAGAAYAIIAAAAYABgAmgAAAA4ADgBYAAAADAAMAGYAAAAQABAAcgAAABAAEACyAAAAFYKI4gYBsR0AAAAPfygaAzoaCa6+fpmNu2/r21AAQQBDAEkARgBJAEMAZABqAG8AaABuAHMASABBAE0AVgBFAFQAMAAyAOAyYXzujroFAAAAAAAAAAAAAAAAAAAAAFQceVB4lD50mtMNAiSke4gMxp1YqbAVMGv56F/9kmCz3UFpS+lKfgo='
> from squid (length: 263).
> 2013/10/10 14:02:11| negotiate_wrapper: Decode
> 'TlRMTVNTUAADAAAAGAAYAIIAAAAYABgAmgAAAA4ADgBYAAAADAAMAGYAAAAQABAAcgAAABAAEACyAAAAFYKI4gYBsR0AAAAPfygaAzoaCa6+fpmNu2/r21AAQQBDAEkARgBJAEMAZABqAG8AaABuAHMASABBAE0AVgBFAFQAMAAyAOAyYXzujroFAAAAAAAAAAAAAAAAAAAAAFQceVB4lD50mtMNAiSke4gMxp1YqbAVMGv56F/9kmCz3UFpS+lKfgo='
> (decoded length: 194).
> 2013/10/10 14:02:11| negotiate_wrapper: received type 3 NTLM token
> 2013/10/10 14:02:11| negotiate_wrapper: Return 'NA =
> NT_STATUS_INVALID_PARAMETER'
>
> The only difference I see there is that negotiate_wrapper gives the
> prefix "KK" for the successful response, but "YR" for the failed one.
> Is it possible negotiate_wrapper is incorrectly classifying the
> response?
This explains the KK/YR and rest of the helper protocol you are seeing:
http://wiki.squid-cache.org/Features/AddonHelpers#Negotiate_and_NTLM_Scheme
YR is given when a user sends a new connection without credentials, or
if limits areound the re-use of an existing connection token have been
exceeded. Normally Type 1 tokens being used to generate Type 2 tokens
(from Squid to client).
KK is given when a Negotiate/NTLM handshake is being completed, or on
Negotiate/Kerberos handshakes. Normally Type 3 tokens being validated.
> I've Googled the crap out of this and haven't found anything useful.
>
> Any suggestions of things I can test, please let me know, I've
> completely run out of ideas.
I suggest upgrading your Squid to the 3.3 package available from the sid
repository. That can be used on Debian 7 without much trouble and will
solve a lot of issues in the 3.1 versions older Debian have.
> Thanks.
>
> squid.conf:
> ------------------------------
> ----------------
> visible_hostname bacoon.pacific.local
> err_html_text http://serviceplus/CAisd/pdmweb.exe
>
> ### negotiate kerberos and ntlm authentication
> auth_param negotiate program /usr/local/bin/negotiate_wrapper -d
> --ntlm /usr/bin/ntlm_auth --diagnostics
> --helper-protocol=squid-2.5-ntlmssp --domain=PACIFIC --kerberos
> /usr/lib/squid3/squid_kerb_auth -d -s GSS_C_NO_NAME
> auth_param negotiate children 256
> auth_param negotiate keep_alive off
>
> ### pure ntlm authentication
> auth_param ntlm program /usr/bin/ntlm_auth --diagnostics
> --helper-protocol=squid-2.5-ntlmssp --domain=PACIFIC
> auth_param ntlm children 256
> auth_param ntlm keep_alive off
>
> ### provide basic authentication via ldap for clients not
> authenticated via kerberos/ntlm
> auth_param basic program /usr/lib/squid3/squid_ldap_auth -R -b
> "dc=pacific,dc=local" -D bacoon-squid_at_pacific.local -W
> /etc/squid3/ldappass.txt -f sAMAccountName=%s -h hector.pacific.local
> auth_param basic children 256
> auth_param basic realm Internet Proxy
> auth_param basic credentialsttl 5 minute
>
> ### ldap authorisation
> external_acl_type memberof %LOGIN /usr/lib/squid3/squid_ldap_group -R
> -K -b "dc=pacific,dc=local" -D bacoon-squid_at_pacific.local -W
> /etc/squid3/ldappass.txt -f
> "(&(objectclass=person)(sAMAccountName=%v)(memberof=cn=%g,ou=Security,ou=Groups,dc=pacific,dc=local))"
> -h hector.pacific.local
> external_acl_type member2 %LOGIN /usr/lib/squid3/squid_ldap_group -R
> -K -b "dc=pacific,dc=local" -D bacoon-squid_at_pacific.local -W
> /etc/squid3/ldappass.txt -f
> "(&(objectclass=person)(sAMAccountName=%v)(memberof=cn=%g,ou=Security,ou=Groups,dc=pacific,dc=local))"
> -h hector.pacific.local
These two external_acl_type have identical parameters. You can drop one
and use the other in its place in all "acl" line definitions.
> ### Use Goblin as upstream cache, don't cache locally
> cache_peer goblin.pacific.local parent 3128 3130 default
> cache deny all
>
> ### acl for proxy auth and ldap authorizations
> acl auth proxy_auth REQUIRED
> # aclname acltype typename activedirectorygroup
> acl RestrictedAccess external memberof "/etc/squid3/access_restricted"
> acl StandardAccess external memberof "/etc/squid3/access_standard"
> acl FullAccess external memberof "/etc/squid3/access_full"
> acl DomAdmins external member2 "/etc/squid3/access_admins"
>
> acl whitelist_noauth_dom dstdomain
> "/etc/squid3/whitelist_noauth_domains"
> acl whitelist_noauth_url url_regex -i
> "/etc/squid3/whitelist_noauth_urls"
> acl whitelist_standard_dom dstdomain
> "/etc/squid3/whitelist_standard_domains"
> acl whitelist_standard_url url_regex -i
> "/etc/squid3/whitelist_standard_urls"
> acl whitelist_restricted_dom dstdomain
> "/etc/squid3/whitelist_restricted_domains"
> acl whitelist_restricted_url url_regex -i
> "/etc/squid3/whitelist_restricted_urls"
> acl whitelist_full_dom dstdomain
> "/etc/squid3/whitelist_full_domains"
> acl whitelist_full_url url_regex -i
> "/etc/squid3/whitelist_full_urls"
>
> acl StreamingRequest req_mime_type -i
> "/etc/squid3/acl_streaming_mimetypes"
> acl StreamingReply rep_mime_type -i
> "/etc/squid3/acl_streaming_mimetypes"
>
> ### squid defaults
> acl manager proto cache_object
> acl localhost src 127.0.0.1/32 ::1
> acl localstuff-nets dst 172.31.0.0/16
> acl localstuff-domains dstdomain .pacific.local
> acl localstuff-domains dstdomain .hsp.local
> acl management src 172.16.0.0/12
> acl management src 127.0.0.1/32 ::1
> acl to_localhost dst 127.0.0.0/8 0.0.0.0/32 ::1
> acl SSL_ports port 443
> acl Safe_ports port 80 # http
> acl Safe_ports port 21 # ftp
> acl Safe_ports port 443 # https
> acl Safe_ports port 70 # gopher
> acl Safe_ports port 210 # wais
> acl Safe_ports port 1025-65535 # unregistered ports
> acl Safe_ports port 280 # http-mgmt
> acl Safe_ports port 488 # gss-http
> acl Safe_ports port 591 # filemaker
> acl Safe_ports port 777 # multiling http
> acl CONNECT method CONNECT
> http_access allow manager localhost
> http_access deny manager
> http_access deny !Safe_ports
> http_access deny CONNECT !SSL_ports
> http_access allow localhost
>
> ### http_access rules
> # enforce authentication, order of rules is important for authorization levels
> http_access allow whitelist_noauth_dom
> http_access allow whitelist_noauth_url
> # Emergency allow all when things get broken:
> #http_access allow all
> http_access deny !auth
> http_access allow FullAccess
> http_access allow DomAdmins
> http_access deny StreamingRequest all
Streaming Request is a simple ACL. It does not need to use the "all"
hack to suppress any special handling.
http_access deny StreamingRequest
> http_access allow RestrictedAccess whitelist_restricted_dom
> http_access allow RestrictedAccess whitelist_restricted_url
> # Standard users can see everything restricted users can see, + more
> http_access allow StandardAccess whitelist_restricted_dom
> http_access allow StandardAccess whitelist_restricted_url
> http_access allow StandardAccess whitelist_standard_dom
> http_access allow StandardAccess whitelist_standard_url
NP: the whitelist_* ACLs are slightly slow, but not as slow as the
external lokup types. If you reverse the order of these ACLs on each
line you can reduce the load on your backend authentication systems.
eg:
http_access allow whitelist_restricted_dom RestrictedAccess
http_access allow whitelist_restricted_url RestrictedAccess
...
> # DO NOT REMOVE THE FOLLOWING LINE
> http_access deny all
>
> http_reply_access deny RestrictedAccess StreamingReply all
> http_reply_access deny StandardAccess StreamingReply all
You can simplify and slightly improve speed by removing the "all" on
above http_reply_access lines.
StreamingReply is a non-auth ACL type. By existing after the
auth-related ACL test it performs the same suppression as the "all" hack
on these lines is designed to do.
Also, what is supposed to happen for all those whitelist_noauth_dom and
whitelist_noauth_url transactions?
they may very well match these reply rules, and *will* be forced to
authenticate before reply delivery due to the *Acces ACL tests being run.
I think you possibly want to add these at the top of your
http_reply_access list:
http_reply_access allow whitelist_noauth_dom
http_reply_access allow whitelist_noauth_url
> ### Always use goblin, except for local stuff
> never_direct deny localstuff-nets
> never_direct deny localstuff-domains
> never_direct allow all
>
> ### SNMP
> acl snmp_public snmp_community public
> snmp_port 3401
> snmp_access allow snmp_public management
> snmp_access deny all
>
> ### logging
> # don't log allowedsites, prioritysites, AnonymousAccess
> #access_log /var/log/squid3/access.log squid !allowedsites
> !prioritysites !AnonymousAccess
>
> ### SquidGuard
> url_rewrite_program /usr/bin/squidGuard
> url_rewrite_children 50
> url_rewrite_access deny localstuff-nets
> url_rewrite_access deny localstuff-domains
> url_rewrite_access deny whitelist_restricted_dom
> url_rewrite_access deny whitelist_restricted_url
> url_rewrite_access deny whitelist_standard_dom
> url_rewrite_access deny whitelist_standard_url
> url_rewrite_access deny whitelist_full_dom
> url_rewrite_access deny whitelist_full_url
> url_rewrite_access deny DomAdmins
> url_rewrite_access allow FullAccess
> # If you're wondering why only FullAccess is allowed above, noone
> except FullAccess should even gete to SquidGuard. The default is allow
> anyway.
I'm wondering why you still have squidGuard?
What is it doing that cannot be done in Squid ACLs?
Amos
Received on Thu Oct 10 2013 - 05:16:02 MDT
This archive was generated by hypermail 2.2.0 : Thu Oct 10 2013 - 12:00:05 MDT