All content types are being indexed. [#337500]

(Nov. 8 dev snapshot)

The sphinxsearch_xmlpipe.php script seems to index all node types regardless of what node types I check in the Sphinx search settings page.

Running sphinxsearch_xmlpipe.php from the command line, I see all of the content returned.

Since some nodes are not meant to be displayed without the proper context, this results in PHP errors or drupal_not_found() errors, both which break the HTML/XML formatting. Consequently, there is no way to index the site.

Comments

Comment #1

mbutcher CreditAttribution: mbutcher commented 21 November 2008 at 20:56

Eventually I got this working. I don't know if it was a cache that didn't get cleared right away (and thus it eventually fixed itself) or the fact that I went into the variables table and manually set node type flags.

Comment #2

markus_petrux CreditAttribution: markus_petrux commented 22 November 2008 at 08:17

Please, go to Administration -> Site configuration -> Sphinx search.

Look at the section "Sphinx index settings". Here you'll find collapsible fieldsets that allow you to select which content types you wish to index (all by default), or which taxonomies you wish to index.

These settings required a rebuild of all indexes, and may affect faceted search block and advanced search form options.

Please note that a site may have a lot of different content types and taxonomies enabled, and it all can turn into something really complex to diggest for the end-user. That's one of the reasons I'm in the process to rewrite a few things about it. Will commit them when I feel it's stable enough.

...or the fact that I went into the variables table and manually set node type flags.

You should clear the corresponding cache item when you do that. You may wish to look at Cache browser module. :)

Comment #3

mbutcher CreditAttribution: mbutcher commented 24 November 2008 at 15:43

I did do that. The problem was, I accidentally (initially) indexed a content type that should not be indexed. When I went back and unchecked it, the "unchecking" did not have any immediate effect. When I ran the xmlpipe generator, I could still see content from the unselected content types.

I then cleared the cache (using the Dev module's cache clearing link). That didn't appear to have any impact.

So I removed the sphinx content type items from the variables table. Then I got sidetracked. When I came back an hour or so later, the xmlpipe PHP page was generating the correct XML output.

I hope that makes my problem clearer.

In the end, though, I don't know whether it was my removing the variables, or simply some cache clearing, that ultimately led to the search server's correct behavior.