orphan Page_Nodes showing up in search

Posted by Community Admin on 03-Aug-2018 17:43

orphan Page_Nodes showing up in search

All Replies

Posted by Community Admin on 20-Apr-2011 00:00

My search results are turning up pages that should no longer exist.  These pages are either deleted older versions of current pages, or pages that simply do not exist anymore.  

A look through the sf_page_node table shows multiple records that simply should not be there -- several records with duplicate titles, and records with titles of pages that no longer exist, which seems to correspond to the errant search results.  I believe this to be a remnant of a mass cleaning/restructuring I've performed on the site structure beforehand.  However, expected behavior is that when you delete a page in the Pages section, the corresponding page_node record should also be removed from the database.  

At any rate, is it safe for me to clear out rogue records from this table?  Other than dependent tables, are there other tables that I might need to sift through as well?  

Thank you

Posted by Community Admin on 20-Apr-2011 00:00

I would actually hope that is not the case if only to allow me to "undelete" which has been high on my wishlist.

If you use the FluentAPI or Managers can you see those pages still in code?

Posted by Community Admin on 20-Apr-2011 00:00

I do see the pages in code when I do a  PagesFacade query = App.WorkWith().Pages().Take(_count_);

However as this seems to query Page_Nodes, this isn't very illuminating.  

To give a little more insight into the issue, I have an instance of a page.  Let's call it Home.  There are 8 page_nodes with url_name_ = 'home'.   

- 3 of these are published.  2 of these published nodes are identical except for voa_version, ids, and create date.  The third is a spanish localized version of this page.

- the remaining 5 have null id2's, null approval_work_state_'s (no 'published' state, for any language).  2 of these have a spanish url (url_name_es), and 1 has a french url (url_name_fr)


I'm pretty sure this mess came about from localization, when I changed the Multilingual URLs mode from "different domains" to "directories.  But right now I'm just concerned about cleaning everything up and potential pitfalls of simply removing the offending entries from the database.  

Posted by Community Admin on 26-Apr-2011 00:00

Hello,

The search does not make database queries to give you some results. We use Lucene that has its own segment files where the data is stored. Can you to recreate/reindex the website and see whether you will keep getting results to old data. Note that currently the search pipe does not allow you to split results by culture  and this is a feature that we are going to have Q2/Q3.

Greetings,
Ivan Dimitrov
the Telerik team


This thread is closed