We are running in to an issue where we get stuck until we go OOM due to some bad handling when the classloader provides URLs such as http://something/.
The issue comes from the fact that the code gets all the URLs the current classloader provides but doesn't seem to handle them properly knowing that it is a URLClassLoader that very well might provide http:// based URLs.

In our case our classloader will provide a url like http://something.something:0/. This is then passed through the info.magnolia.cms.util.ClasspathResourcesUtil.sanitizeToFile(URL) method which will strip this down to /.

The result of this is that the info.magnolia.cms.util.ClasspathResourcesUtil.collectFiles(Collection<String>, File, Filter) method will collect every file on the whole system.