Question

I admit I’ve cobbled together a mostly working production setup on Ubuntu with Capistrano from the official docs (which seem dated and make a lot of assumptions) and various blog posts of varying outdatedness. Anyway, the last annoying hang up is that indexing works when I do it by hand (and on deploy I’m pretty sure), but doesn’t work from Cron.

Here’s my crontab:

$ crontab -l
# m h  dom mon dow   command
* * * * * cd /var/www/app/current && /usr/local/bin/rake RAILS_ENV=production thinking_sphinx:index >> /var/www/app/current/log/cron.log 2>&1

Here is the log output (this actually appears 3 times per call):

Sphinx cannot be found on your system. You may need to configure the following
settings in your config/sphinx.yml file:
  * bin_path
  * searchd_binary_name
  * indexer_binary_name

For more information, read the documentation:
http://freelancing-god.github.com/ts/en/advanced_config.html

This is when I run the same command by hand (also works when logging):

$ cd /var/www/app/current && /usr/local/bin/rake RAILS_ENV=production thinking_sphinx:index
(in /var/www/app/releases/20100729042739)
Generating Configuration to /var/www/app/releases/20100729042739/config/production.sphinx.conf
Sphinx 0.9.9-release (r2117)
Copyright (c) 2001-2009, Andrew Aksyonoff

using config file '/var/www/app/releases/20100729042739/config/production.sphinx.conf'...
indexing index 'app_core'...
collected 5218 docs, 3.9 MB
collected 5218 attr values
sorted 0.0 Mvalues, 100.0% done
sorted 0.7 Mhits, 100.0% done
total 5218 docs, 3898744 bytes
total 0.616 sec, 6328760 bytes/sec, 8470.28 docs/sec
distributed index 'app' can not be directly indexed; skipping.
total 3 reads, 0.008 sec, 1110.2 kb/call avg, 2.6 msec/call avg
total 15 writes, 0.016 sec, 540.4 kb/call avg, 1.0 msec/call avg
rotating indices: succesfully sent SIGHUP to searchd (pid=20101).

Also relevant:

$ which rake        
/usr/local/bin/rake

$ which indexer
/usr/local/bin/indexer

The error is somewhat common, but it smells funny that it works fine from the command line, I suspect something else is weird. I have 2 other mission-critical cron jobs that run rake tasks that look identical and run fine, not sure what's different about this one. Any help would be greatly appreciated!

PS-is there an authoritative deploy config for this with current Capistrano and TS versions? It seems everyone rolls their own, and the official docs seem to be as idiosyncratic as the blog posts out there.

Answer 1

Is the crontab owned by the same user as the one you are logged in as when you run things manually?

Since it seems like a clear PATH issue, and cron runs with a restricted PATH (i.e. not what’s in your .profile), try adding this to the top of your crontab file.

PATH=/bin:/sbin:/usr/bin:/usr/sbin:/usr/local/bin

Or if you don’t want to modify cron’s PATH, you could symlink the files you need into /usr/sbin, which is likely in the PATH by default.