On 24/12/10 01:45, Alexander Weber wrote:
Dear Squid Gurus!
I am running squid 2.6 on an IPCOP firewall as caching proxy for my internal
network.
Well, first of all an upgrade would be worthwhile if you can. 2.6 is
very old now.
I got 2 more servers now running an HTTPS service. Our webhoster forwards
traffic by URL-HIDING to our dynDNS.
Please describe this "URL-HIDING" concept. Were there any technical
terms they used to describe it?
a) internal clients being able to browse the web via squid
b) HTTP website on server APP-001 available at port 80 and URL xxx.mine.nu,
cached to the outside
c) HTTPS website on server NAS-003 available at port 443 and URL
xxx.mine.nu, cached to the outside
d) HTTPS website on server APP-004 available at port 443 and URL
xxx.mine.nu/extranet/, cached to the outside
NOTE: Squid will happily multiplex traffic. Requests on port 80 to
backend on port 443 and vice-versa if allowed to.
ALSO NOTE: (d) is a sub-set of (c) and handling them in those order will
prevent the (d) being done on most requests.
I see that your config below does not match these requirements. In
that it extends (c) slightly by adding config equivalent to "except for
xxx.mine.nu/extranet/"
In the end it should work like this:
Web proxy:
internet<---- squid transparent:800<---- internal clients browsing the web
Reverse proxy:
internet (xxx.mine.nu) ----> squid accel:80 ---->
APP-001:80
internet (xxx.mine.nu) ----> squid accel:443 ---->
NAS-003:443
internet (xxx.mine.nu/extranet/) ----> squid accel:443 ---->
APP-004:443 (or :80)
Based on my squid.conf I am able now to grab the standard webpage
(APP-001:80), the /extranet/ page for APP-004 at port 80 and 443 (which
should be only 443) but the webpage at 443 is not opening. Furthermore every
Please explain "the webpage at 443 is not opening" this does not sound
right. From where is it not opening; direct request to the peer? or only
wen relayed through squid?
traffic from internal network is ending now on the webserver at APP-001:80.
Problem #0:
That sounds a lot like you are NAT'ing the LAN traffic to port 80
instead of to port 800.
Here is my squid.conf:
[CODE]
shutdown_lifetime 1 seconds
icp_port 0
# Reverse Proxy for HTTP
http_port 80 accel defaultsite=www.REALDNS.net
# Forward ORIGIN IP to server - to be checked
#forwarded_for on
cache_peer 192.168.103.25 parent 80 0 no-query originserver name=app-001
acl ndjsite dstdomain www.REAL.net
cache_peer_access app-001 allow ndjsite
# Deny - check later
#cache_peer_access app-001 deny all
Problem #1:
all traffic arriving at "http_port 80" the defaultsite= option will
generate URLs as requests for:
http://www.REALDNS.net/*
in order to match "www.REAL.net" you will need to use
"defaultsite=www.REAL.net" or more likely add the "vhost" option.
Also missing the http_access rule matching your cache_peer_access lines
to permit outside access to this site. These are mandatory for public
reverse-proxy.
http_access allow ndjsite
NP: the cache-peer_access also breaks your (b) requirement that
"xxx.mine.nu" be the domain served by this peer. Since that domain is
not permitted to be fetched there.
# w/ cert for HTTPS
https_port 443 accel cert=/var/ipcop/proxy/server.crt
key=/var/ipcop/proxy/server.key defaultsite=xxx.mine.nu vhost protocol=https
# Forward ORIGIN IP to server - to be checked
#forwarded_for on
# and multi host
acl ndjext urlpath_regex ^/extranet
cache_peer 192.168.103.20 parent 443 0 no-query originserver ssl
name=nas-003 login=PASS
cache_peer_access nas-003 deny ndjext
Possible problem #2:
implicit "nas-003 allow all" end this sequence of tests.
Its good that you had this "deny ndjext" which will allow the (d)
requirement to happen later. However you still need to follow it here
with an "allow ..." line permitting the (c) requirement domain to be
sent here
I think you want:
acl xmu dstdomain xxx.mine.nu
cache_peer_access nas-003 allow xmu !ndjext
cache_peer_access nas-003 deny all
http_access allow xmu
cache_peer 192.168.103.26 parent 443 0 no-query originserver ssl
name=app-004 login=PASS
cache_peer_access app-004 allow ndjext
cache_peer_access app-004 deny ndjsite
Problem #3,4,5:
- "allow ndjext" permits *any* website using the text "/extranet" at
start the path to be sent to this peer.
- again the list ends with an implicit "allow all" permitting
everything to be sent here. (except the ndjsite which is denied)
I think you want:
cache_peer_access app-004 allow xmu ndjext
cache_peer_access app-004 deny all
http_port 192.168.103.1:800 transparent
http_port 192.168.105.1:800 transparent
acl QUERY urlpath_regex cgi-bin \?
no_cache deny QUERY
cache_effective_user squid
cache_effective_group squid
pid_filename /var/run/squid.pid
cache_access_log /var/log/squid/access.log
cache_log /var/log/squid/cache.log
cache_store_log none
log_mime_hdrs off
forwarded_for off
# Do not modify '/var/ipcop/proxy/squid.conf' directly since any changes
# you make will be overwritten whenever you resave proxy settings using the
# web interface! Instead, modify the file '/var/ipcop/proxy/acl' and then
# restart squid using the web interface. Changes made to the 'acl' file
# will propagate to the 'squid.conf' file at that time.
# [Scott Tregear, 22 Feb 2005]
# Uncomment the following line to enable logging of User-Agent header:
#useragent_log /var/log/squid/user_agent.log
# Uncomment the following line to enable logging of Referer header:
#referer_log /var/log/squid/referer.log
acl all src 0.0.0.0/0.0.0.0
acl localhost src 127.0.0.1/255.255.255.255
acl SSL_ports port 443 563
acl Safe_ports port 80 # http
acl Safe_ports port 21 # ftp
acl Safe_ports port 443 563 # https, snews
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 Safe_ports port 800 # Squid port (for icons)
acl IPCop_http port 81
acl IPCop_https port 445
acl IPCop_ips dst 192.168.103.1 192.168.105.1
acl IPCop_networks src 192.168.103.0/255.255.255.0
192.168.105.0/255.255.255.0
acl no_proxy_dst dst 0/32
Bit weird that ACL. We write it in the current default config as:
acl to_localhost dst 127.0.0.1 0.0.0.0/32
acl CONNECT method CONNECT
##Access to squid:
#local machine, no restriction
http_access allow localhost
#GUI admin if local machine connects
http_access allow IPCop_ips IPCop_networks IPCop_http
http_access allow CONNECT IPCop_ips IPCop_networks IPCop_https
#Deny not web services
http_access deny !Safe_ports
http_access deny CONNECT !SSL_ports
#Finally allow IPCop_networks clients
http_access allow IPCop_networks !no_proxy_dst
http_access deny all
maximum_object_size 64768 KB
minimum_object_size 4 KB
cache_mem 10240 KB
cache_dir aufs /var/log/cache 256 16 256
request_body_max_size 0 KB
reply_body_max_size 0 allow all
visible_hostname loki.xxxxxxx.local
[/CODE]
My config base was the squid manpages, squidcache.org's config examples and
these page:
http://www.sweetnam.eu/index.php/Reverse_Proxy_with_Squid and
http://www.techienuggets.com/Comments?tx=94569
Perhaps you can point me to some examples were a setup like mine is running
or where I can look for a solution. Google, wiki.squid-cache.org and these
pages are my friends already:
http://blog.spench.net/2010/02/24/tips-for-setting-up-squid-in-reverse-proxy
-web-accelerator-accel-mode/
Hmm, couple of quite nasty assumptions there. Firstly about how to run
Apache and Squid on the same machine. Then a lot of FUD about "tighten
things up".
http://www.linux-club.de/viewtopic.php?t=24381
This one appears to be relevant for 2.5. Reverse proxy and multiple
modes were VERY different in that version and not easy.
Amos
--
Please be using
Current Stable Squid 2.7.STABLE9 or 3.1.9
Beta testers wanted for 3.2.0.3
