A couple of days ago one of our customers opened a ticket with us regarding a failure with the Link Validation scheduled job on one of their websites. The history tab of the job reported the following message: this operation is not supported for a relative URI. A quick Google on the message pointed me to a similar issue on World from 2015.
In the post, they mentioned a query to get a list of broken links. We ran the query and got 23 results, not that many luckily. Most of the broken links were relative paths, just like what the message indicated, but there were a few local file references mixed in too and two that started plainly with a letter. Now we consider all the results from the query to be broken links.
Now, why is it that the Link Validator, who has the job of finding broken links, fails when it finds broken links?!
Yes, the post also mentions a way to do filters (excludePatterns). But then we are just ignoring the broken links altogether aren't we? It makes sense for exclude for tel, but excluding for file:// or relative paths is the last thing we want.
We consider this behaviour to be a bug on Episervers side.
After having almost pressed the submit button I noticed that the job was done, after 99 minutes and 37 seconds it failed again, this time with the message "Invalid URI: The Authority/Host could not be parsed.". A common exception message that you can get when you try and use a URL like "http:/www.google.com/" (note the missing slash). A little bit of querying gave me a result where this is indeed the case.... I don't think I can ever recommend this Link Validator to any of our customers in this state.