Apache mod_proxy_balancer: No Protocol handler was valid

Just this afternoon, one of my colleagues and I were discussing our feelings about the “semicolon-class” of bugs that developers will inevitably spin their wheels on from time to time. I’ve had just such an experience the past two evenings with what started out as a simple recipe from the Apache Cookbook, entitled “Load Balancing with mod_proxy_balancer”.

The recipe itself is straightforward, taking up just over a page. I was able to get a basic load balancer working within a few minutes by following the recipe, but for some reason I couldn’t get the balancer-manager site to load correctly. The following directives should configure the balancer-manager:

<Location /balancer-manager>
  SetHandler balancer-manager
  Order Deny,Allow
  Allow from all
</Location>

When I would start the server, however, I got an HTTP 500 error response, with this message in the log:

[Tue Feb 02 12:49:09 2010] [warn] proxy: No protocol handler was valid for the URL /balancer-manager. If you are using a DSO version of mod_proxy, make sure the proxy submodules are included in the configuration using LoadModule.

My searches on Google were fruitless: most people were able to solve the problem by ensuring these lines were added to their httpd.conf:

LoadModule proxy_module modules/mod_proxy.so
LoadModule proxy_balancer_module modules/mod_proxy_balancer.so
LoadModule proxy_ftp_module modules/mod_proxy_ftp.so
LoadModule proxy_http_module modules/mod_proxy_http.so
LoadModule proxy_connect_module modules/mod_proxy_connect.so

But they were already included in my conf file, so no dice. Other people solved their problem by loading the mod_proxy_html.so, which doesn’t come on a standard install of Apache. I ended up compiling it from source (learned a lot about compiling Apache shared modules) but that didn’t work for me either.

Eventually I realized that the only time my requests were being proxied to the BalancerMembers was when I was at the URL root, for example tacking on an ‘index.php’ to the URL would generate the same error as trying to access ‘balancer-manager’. And about that time I stumbled upon a user who was able to solve their problem by adding a few rewrite rules… which caused me to finally understand part of my problem.

The reason ‘balancer-manager’ wasn’t loading correctly was because that request was being proxied to the BalancerMember, which didn’t have ‘balancer-manager’ enabled (hence the ‘no protocol handler error’). This still didn’t explain why ‘index.php’ wasn’t working, either, since I knew index.php was on the BalancerMembers.

These discoveries quickly culminated into leading me to my solution, which was a missing trailing-slash in my ProxyPass directive. That’s what was causing my ‘index.php’ to not be proxied. Once I realized the problem was in my ProxyPass– reading the documentation also taught me that there’s a way to keep certain requests from being proxied (which solved my ‘balancer-manager’ issue).

So I present to you, my completed (and functional) balancer.conf. And remember, the trailing slash at the end of the ProxyPass directive IS CRUCIAL!

ProxyPass /balancer-manager !  # Prevents balancer-manager from being proxied.
ProxyPass / balancer://mycluster/
ProxyPassReverse / balancer://mycluster/

<Proxy balancer://mycluster>
  BalancerMember http://10.0.0.10 loadfactor=2
  BalancerMember http://10.0.0.11
  BalancerMember http://10.0.0.12
</Proxy>

<Location /balancer-manager>
  SetHandler balancer-manager
  Order Allow,Deny
  Allow from all
</Location>

I hope this saves someone some time when searching on the particular error message I got. Now I’m off to figure out how to get my web servers to auto register with the balancer-manager. I will blog about it when I figure it out.