Closed (fixed)
Project:
Nutch search engine integration
Version:
6.x-1.x-dev
Component:
Code
Priority:
Normal
Category:
Feature request
Assigned:
Reporter:
Created:
6 Nov 2010 at 17:47 UTC
Updated:
17 Nov 2010 at 05:08 UTC
Jump to comment: Most recent file
Comments
Comment #1
dstuart commentedUseful will add it to trunk today
Comment #2
robertdouglass commentedThis should really be 3-4 patches, but I'll stick it here and see if you complain ;-)
Aside from several niggly nitpicks about whitespace, and a tiny typo correction in a description, this patch shuffles some of the variable setting and getting around to make a little clearer distinction between nutch_admin_settings and nutch_admin_crawl. Also, I think that some of the nutch_admin_settings variables were not being set correctly, so that bug is fixed as well.
Comment #3
robertdouglass commentedOk - I know you're merging things and it's a pain, but until we're on git it's going to be a pain. If you commit what you've got now I'll merge the stuff in from this patch myself.
This is a new improvement (I hope) that streamlines the directory checking for the logs directory. I also had to learn that the is_dir function needs not only the NUCTH/logs dir to be g+wrx, but the NUTCH directory and any shared parents with Apache DOCROOT need g+rx.
So, if my Nutch directory is /home/robert/lib/nutch, here's what stuff has to look like for www-data to be able to do its thing:
/home/robert/Sites/docroot <-- apache docroot
which means:
/home/robert/lib robert:www-data g+rx
/home/robert/lib/nutch robert:www-data g+rx
It wasn't enough that just /home/robert/lib/nutch was writeable with www-data, it needs +x too. Confusing!
Comment #4
robertdouglass commentedSame as above but with one more directory check.
Comment #5
dstuart commentedHave commited merged changes up to #2, plus runbot enhancements
Comment #6
robertdouglass commentedI tried to merge my patch with your commits. Review with care, but it seems to be ok. My goals with this patch were to simplify the reporting when directory permissions are not correct. I think it got a little better, but this is still a big sticking point, as it is rather tricky to set the directory permissions in a way that is maximally secure while still actually allowing Drupal to run the process.
Comment #7
dstuart commentedHey,
Have updated the patch to be inline with trunk.
Regards,
Dave
Comment #8
robertdouglass commentedHere's my current patch against head.
Mostly same as above with two extra lines removed for cleanup (including the line from #971278: drupal_set_message on $process not helpful)
Comment #9
robertdouglass commentedsame as above with some whitespace differences.
Comment #10
dstuart commentedHi,
Have commited patch to head, thanks Robert
Regards,
Dave