Cron reports indexing/rendering errors

Only from Drush. When cron run as scheduled or manually from the Status Report Page, I do not see any errors.

Please let me know if I can send any info to aid in your debugging.

Thanks!

I was incorrect about this error only being thrown when running cron via Drush. Unfortunately, the error seems very intermittent, which probably means I have not yet figured out what triggers it. However, in the past two days, several scheduled cron runs threw these errors. If you have any ideas what I can do to help track this down, it would be much appreciated!

Thanks for this helpful info. My Search by Pages search environment is set to use the "Anonymous user" role for indexing. I did check many of the pages listed and they were all visible to anonymous users.

Just to complicate things, the last couple of cron runs instead have been throwing these errors:

Notice: Trying to get property of non-object in node_page_title() (line 2075 of /path/to/site/modules/node/node.module

I would dearly love to track this down. To summarize, here's what I'm observing:

• Running cron via drush always throws the "WD search_by_page: Content not rendered" error
• Running cron via admin/config/system/cron never throws either error
• Running cron via trigger by anonymous user visit sometimes or always throws "Trying to get property of non-object in node_page_title" error
• Running cron via trigger by logged in user visit sometimes or always throws "WD search_by_page: Content not rendered" error

I am somewhat less certain of the last two points, as it's hard to test them to determine repeatability.

You got me thinking when you said that Search by Pages looks for the main content block. Since running cron via drush always caused the "WD search_by_page: Content not rendered" error, I could finally track it down. Here's what caused that long list of errors -- on *one* page I didn't include the main content block. I'm using contexts to set up my blocks and one page had a slightly different structure, so I didn't include the main content block. Big mistake! As soon as I put that block in on every page, "drush core-cron" runs with no problem at all.

Thank you for your ideas above. That's gotten rid of one bug. Now, hopefully I can figure out the other one.

I've done some further sleuthing and I think that the "Trying to get property of non-object in node_page_title" error has nothing to do with Search by Page. Instead, it seems related to Boost, so I'm reporting the bug in that issue queue. Thanks for all of your help.

I may have been incorrect that Boost is to blame for this bug. I'm sorry to be all over the map on this, but it's been hard to debug. I finally figured out that I could temporarily set my site to run cron whenever I visited a page by changing system_run_automated_cron to always run cron, regardless when it was last run. OK, after the should-have-been obvious move of disabling Boost, I still see the bug. However, when I disable Search by Page Nodes, the bug is gone. To be more specific:

1. Disable Boost
2. Disable Search By Page Nodes
3. Set system_run_automated_cron to always run cron
4. Visit a page on the site (either logged in or anonymous)
5. Log has a bunch of errors of the form "Content was skipped" which I'm assuming (hoping) are not big deal.
6. Enable Search by Page Nodes
7. Visit a page on the site (either logged in or anonymous)
8. Log has a bunch of errors of the form "Trying to get property of non-object in node_page_title".

The stacktrace in node_page_title when the error is thrown is these functions:

node_page_title, call_user_func_array, _menu_item_localize, _menu_link_translate, _menu_tree_check_access, menu_tree_check_access, menu_build_tree, menu_tree_page_data, menu_tree_add_active_path, menu_tree_build, menu_block_block_view, call_user_func_array, module_invoke, _block_render_blocks, block_list, block_get_blocks_by_region, execute, context_page_build, search_by_page_page_content, search_by_page_update_index, call_user_func_array, module_invoke, search_cron, call_user_func_array, module_invoke, drupal_cron_run, system_run_automated_cron, drupal_page_footer, drupal_deliver_html_page, drupal_deliver_page, menu_execute_active_handler

When I remove all menu blocks from the page, I do not get an error. The call to menu_tree_page_data in menu_tree_add_active_path seems to be the offender.

Any thoughts? I plan on cross-posting this in the Menu Block issue queue, as I have no idea in which module the problem is originating.

I also want to flag this post as being related: Improper usage of node_page_title() function by _menu_item_localize() function in includes/menu.inc

Thanks for taking a look at this. I did the sleuthing and here's what I observed:

1. Disable Menu Blocks, run cron via visiting www.example.com/about -- log does not have any errors of the form "Trying to get property of non-object in node_page_title".
2. Via Context, add a regular menu to www.example.com/about, run cron via visiting visiting www.example.com/about -- log does not have any errors of the form "Trying to get property of non-object in node_page_title".
3. Enable Menu Blocks, confirm that no menu blocks are on www.example.com/about, run cron via visiting visiting www.example.com/about -- log does not have any errors of the form "Trying to get property of non-object in node_page_title".
4. Add menu block to www.example.com/about, run cron via visiting visiting www.example.com/about -- log has many errors of the form "Trying to get property of non-object in node_page_title".

The error is initiated by Menu Block's function menu_tree_add_active_path, which calls menu_tree_page_data. When I replace it with a call to menu_tree_all_data, I do not get the error. Both of these functions return the menu tree, but menu_tree_page_data indicates the active trail, which is key to its functionality.

I don't know how this could be a considered a bug in either Search By Page or Menu Block. When Menu Block calls menu_tree_page_data, it uses appropriate arguments and gets a nice result back -- I can't see any problem with the array returned. It's just along the way, the core calls node_page_title improperly. It must have something to do with the spoofing done by Search By Page which you mention, and it'd probably be quicker to fix what's going wrong there than to try to fix core, so I'd like to keep digging.

Can you give me some more info about the spoofing so I can look into that?

Thanks!