To create more complex rules for purging your content, you can add an ECCU file to your request.
The ECCU file, also referred to as the purge.data file, is a text file in a simplified XML format that supports the following tags:
<eccu> </eccu>– obligatory opening and closing tags that indicate the file type.
<revalidate>– specifies a timestamp value that tells the edge servers to revalidate objects with older timestamps before serving them. See Revalidate tag for more information.
<match>– specifies the URI or other attributes that identify the objects you want to revalidate. There are three types of matches you can add to your ECCU request:
The path match that specifies the URL path.
The request property match that identifies the request by an attribute other than the URL, such as query string arguments, file extensions, or
The response property match that identifies the request by an attribute in the response from the origin server, that is the
Mind the limitations
Use these matches with caution. You can create a large number of unique matches and thus build a large data file and adversely affect site performance. See ECCU performance considerations.
Your ECCU file persists, and the code lines with the matches persist until you replace them. Matches are replaced if a new match includes the old one at a parent level. For example, if you previously used
/bin/images/*.gifas a match and now updated the request to include
/bin/*.*, the latter match replaces the former, assuming the timestamp is for a current date and time. Matches are replaced or pruned if there are new ones with more recent timestamps.
Since requests persist in the ECCU file, it grows without bound if not maintained properly. To prevent growth to a point that impacts the performance, the edge server automatically prunes the requests using the
/*.*match (all directories, all extensions) with a timestamp as distant in the past as possible, and consequently reduces the file size.
For better performance and maintenance, keep the match files small and occasionally prune them with a match of
/*.*and a timestamp of, for example, two weeks ago to keep the file size down.
Updated 9 months ago