Apache reverse proxy and load balancer - does not work as it should - apache

Question

I have 3 machines.
One (loadbalance.lan) is used as a load balancer, the other two (172.16.30.5 and 172.16.30.6) are tomcat's servers. Main page of the tomcat is listening on port 8080
Im typing in the browser loadbalance.lan/tomcat and I am able to see one of the tomcat content (default tomcat page)
The problem is page isn't displayed correctly. There's no images and when I click on any link it displays 404 Not found error.
Lets say I want to access one of the sub pages on the tomcat website. Tomcat website address: 172.16.30.5:8080
Now I can choose, lets say "status" link which redirects me to: 172.16.30.5:8080/manager/status (and works fine)
When I access the same page but via reverse proxy server (loadbalance.net) and click that link on the loadbalance.lan page, links redirect me to loadbalance.lan/manager/status and I get 404 error.
Of course when I type in the browser loadbalance.lan/tomcat/manager/status it displays correct.
Problem with the images is also weird. When I use url: loadbalance.lan/tomcat I can't see images (Tomcat logo)
When I use this one: loadbalance.lan/tomcat/ (slash at the end) it's ok. At least images because links still redirect in wrong place.
Here is my loadbalance.lan apache config:
<Proxy *>
Order deny,allow
Allow from all
</Proxy>
<VirtualHost *:80>
ProxyRequests Off
ProxyVia On
ProxyPreserveHost On
<Proxy balancer://cluster>
Order Deny,Allow
Allow from all
</Proxy>
<Proxy balancer://cluster>
BalancerMember http://172.16.30.5:8080
BalancerMember http://172.16.30.6:8080
<Proxy balancer://cluster>
</Proxy>
<Location /tomcat>
ProxyPass balancer://cluster
ProxyPassReverse balancer://cluster
</Location>
</VirtualHost>
Could someone help me with this?
Obviously there is something wrong with that proxy but I have no idea how to fix that :(

Answer 1

From ProxyPassReverse documentation (strong added):
This directive lets Apache adjust the URL in the Location, Content-Location and URI headers on HTTP redirect responses. This is essential when Apache is used as a reverse proxy (or gateway) to avoid by-passing the reverse proxy because of HTTP redirects on the backend servers which stay behind the reverse proxy.
Only the HTTP response headers specifically mentioned above will be rewritten. Apache will not rewrite other response headers, nor will it rewrite URL references inside HTML pages. This means that if the proxied content contains absolute URL references, they will by-pass the proxy. A third-party module that will look inside the HTML and rewrite URL references is Nick Kew's mod_proxy_html.
So, the proxy job is not to rewrite the html content of the pages, if the proxyied content does not know that the final url should contain /tomcat extension and the proxy does not alter the pages... you're stuck.
This is usually something you do not see because the 172.16.30.5:8080 part is well rewritten in localhost.lan, but this rewrite is not made by the proxy, quite certainly because urls are in fact only relative (<img src="/foo/bar.png">). Check the source code of the page to see if the domain name is really rewritten in urls).
There's several ways of handling that:
- You could avoid altering relative urls paths in, the proxy (so not using a tomcat/ prefix, but instead a dedicated virtualhost with a name, like tomcat.lodabalncer.lan).
- You could also use some dedicated tools, like mod_proxy_html to rewrite the content of the pages, but that's a slow and complex thing.
- The third way is to manage the final full url on the application side (here tomcat) and detect the proxy chain elements in X-Forwareded-for Header to rebuild the right domain.
- Some applications provides tools for that, like the VirtualHostMonster in Zope
For tomcat the preferred tool is mod_proxy_ajp and not mod_proxy. But for a load balancer proxy I do not think you can use mod_proxy_ajp. And, it's been a long time since I made this, but in my memory I think mod_jk was the solution to that.
Read this full documentation on tomcat proxying for details. At least you should get some hints for the solution.