The previous post detailed how to configure a High Availability cluster for redundant directors. Paired with the NLB cluster, the framework is there for a resilient system. Now we have to configure services to run inside the cluster - namely Squid (and later, SquidGuard). It is important to keep backend configuration consistent as well, which will be covered in the post following this one.
Most of this post discusses the Squid conf file.
The first thing to do is install Squid on each NLB cluster node (LVS-Cache1, LVS-Cache2 & LVS-Cache3). The package included with Ubuntu works perfectly. The installer will insert a startup script into /etc/init.d too. It installs an example conf file into /etc/squid3 which is very heavily commented, so stop Squid, rename the conf file and download my basic template below.
root@lvs-cache1:~# apt-get update root@lvs-cache1:~# apt-get upgrade root@lvs-cache1:~# apt-get --yes install squid3 ntp curl root@lvs-cache1:~# /etc/init.d/squid3 stop root@lvs-cache1:~# mv /etc/squid3/squid.conf /etc/squid3/squid.conf.dist
Below is each part of my conf file with an explanation. Only create the new conf file on one cluster node as we will configure file synchronisation in the next post. This is actually a very basic conf file so you should modify it as per your requirements. The Squid documentation is very comprehensive so should be your first port of call for more advanced topics such as authentication (although I will cover this in a future post).
Once this initial configuration is complete, we will move on to synchronising it with the rest of the cluster, and then return to it to configure SquidGuard.
Download the basic conf file here. Rename it to /etc/squid3/squid.conf.
Configuration
http_port 3128 cache_mgr admin@mydomain.local maximum_object_size 20480 KB
- Listen on port 3128. If you've specified a different port in the directors' ldirectord.cf file then make sure it's reflected here.
- Email address of the cache administrator who can be contacted if the cache dies.
- Maximum object size - self explanatory 
acl manager proto cache_object acl localhost src 127.0.0.1/32 http_access allow manager localhost http_access deny manager acl localnet src 192.168.0.0/24
- Limit cache_object access to localhost (this is information on how Squid is configured - it's just like an HTTP request but Squid generates the diagnostic data itself).
- Define the localnet acl to include the IP range of LAN clients. This ACL is used below.
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
- Define the range of ports allowed through Squid. In most cases HTTP, HTTPS and FTP are enough but these are the ports defined by default so unless you have a specific reason to prohibit certain ports, it is safe to leave the defaults.
http_access deny !Safe_ports http_access deny CONNECT !SSL_ports
- Only allow HTTP connections from safe ports (deny ports not defined in Safe_ports ACL).
- Only allow SSL connections from SSL_ports (deny ports not defined in SSL_ports).
http_access allow localhost http_access allow localnet http_access deny all
- Allow HTTP access from localhost.
- Allow HTTP access from LAN clients.
- Deny all other HTTP connections.
icp_access deny all htcp_access deny all
- Prohibit all access attempts to cache management protocols.
- Read more on ICP and HTCP.
always_direct allow all
- Allow direct HTTP connections (this only needs changing if you use a parent proxy, in which case direct connections are generally not desired).
hierarchy_stoplist cgi-bin ?
- Always fetch CGI URLs directly (do not query neighbor caches). This line is here as a failsafe for future cache_peer configurations.
access_log /var/log/squid3/access.log squid
- Write a local log of cache accesses. This is a temporary measure until the final part of this series which concentrates on centralised logging.
refresh_pattern ^ftp: 1440 20% 10080 refresh_pattern ^gopher: 1440 0% 1440 refresh_pattern (cgi-bin|\?) 0 0% 0 refresh_pattern . 0 20% 4320
- Set expiry times for objects that do not have an explicit expiry time - these are the defaults.
- http://www.squid-cache.org/Doc/config/refresh_pattern/
coredump_dir /var/spool/squid3 shutdown_lifetime 5 seconds
- Set the directory Squid leaves coredump files (in the even the daemon crashes).
- Wait 5 seconds for HTTP connections to close. Default is 30 seconds.
Testing
Test the configuration. If the result is a single line then all is well:
root@lvs-cache1:/etc/squid3# squid3 -k parse 2011/08/04 02:35:46| Processing Configuration File: /etc/squid3/squid.conf (depth 0) root@lvs-cache1:/etc/squid3#
If not, you should get an indication of where the error in the file lies:
root@lvs-cache1:/etc/squid3# squid3 -k parse 2011/08/04 02:35:46| Processing Configuration File: /etc/squid3/squid.conf (depth 0) 2011/08/04 02:35:46| cache_cf.cc(363) parseOneConfigFile: squid.conf:2 unrecognized: 'zhttp_port' root@lvs-cache1:/etc/squid3#
Once you have a working conf file, go ahead and start Squid. Observe the cache.log file for any additional errors.
root@lvs-cache1:~# /etc/init.d/squid3 start
Check with ipvsadm on the active director - the first cache server should have a weight of 1 with no connections:
root@LVSTest-Director1:~# ipvsadm -ln IP Virtual Server version 1.2.1 (size=4096) Prot LocalAddress:Port Scheduler Flags -> RemoteAddress:Port Forward Weight ActiveConn InActConn TCP 192.168.150.210:3128 wlc -> 10.2.0.1:3128 Route 1 0 0 -> 10.2.0.2:3128 Route 0 0 0 -> 10.2.0.3:3128 Route 0 0 0
The last task in this part is to actually test the proxy. Configure a client that you have allowed access in Squid's conf file and try loading a webpage. If all goes to plan, the request should be processed straight away. Check ipvsadm again - notice the ActiveConn count has gone up. You can also double check by looking at the access log file (if you have one configured) on the cache box itself.
root@LVSTest-Director1:~# ipvsadm -ln IP Virtual Server version 1.2.1 (size=4096) Prot LocalAddress:Port Scheduler Flags -> RemoteAddress:Port Forward Weight ActiveConn InActConn TCP 192.168.150.210:3128 wlc -> 10.2.0.1:3128 Route 1 3 0 -> 10.2.0.2:3128 Route 0 0 0 -> 10.2.0.3:3128 Route 0 0 0
You are now ready to move on to the next post - synchronising configuration.