The linkcheck cronjob verifies if external urls are valid using eZHTTPTool.

As no result data needs to be returned, curl is set the option to only get headers, and thus performs a HEAD request.

However, some sites do not respond in the same manner as to a GET request, and return error status codes (http 500, 503, 405).

A few examples:

$ curl --head http://www.wn-immo.de/search.html
HTTP/1.1 500 Internal Server Error

$ curl --head http://www.amazon.com/
HTTP/1.1 405 MethodNotAllowed

For compatibility, curl should use a standard GET, while still only downloading headers.