trouble shooting

In case of a problem, please contact:

  • Outhink Support: support at outhink dot com
  • Markus Sandy: markus at outhink dot com (skype msandy)
  • JD Lasica: jdlasica at gmail dot com

Some things that have happened before and what we did about it:

  1. user 0 record is missing from database - INSERT a new one into the users table
  2. out of disk space - purge database logs as needed
  3. request flood from a specific IP address - block address
  4. modules screen is white - remove unnecessary modules, increase memory allocation
  5. users can't log in - clear browser cookies may fix for user, but see Drupal.org for server-based solutions. Much discussed issue on Drupal.org
  6. system slowdown - purge sessions and watchdog tables as needed

I have been manually cleaning out the sessions table for some time now. We get about 8000 records per hour and so it does not take long to reach several hundred thousand records. Things start to bog down then. The system becomes very slow, especially if we are displaying the "Who's online" block on the home page, which is something we like to do. Also, the query to delete old session records can take quite a while to perform (up to 18 minutes that I have seen).

JD indicated that he could no longer upload files under one of his accounts. I figured it was a quota somewhere. I googled "upload drupal quota" and found this post on drupal.org. It turns out that upload quotas are set by role. How nice. I increased the upload quota for site admins from 300MB to 500MB total for now. Will look into cleaning up eventually. This is in /admin/settings/upload (4.6) in case we need to bump again. Currently, max size per upload is set to 1MB for site and media admin roles.

Down again.

By Markus Sandy at 05/05/2007 - 09:48

Eric pinged me yesterday afternoon to say that Ourmedia was down. I logged into the webserver using secure shell and it was crawling. So slooooooooooow. There were 92 httpd processes according to

Ourmedia went blank the other day for a few moments. Apache was up. MySQL was up. There had been no recent changes to code.

The user 0 record was missing again. It is a mystery as to what causes this. I had just dropped a database, but that should not have had any effect.

Luckily it was a quick restore of a record to the users table. This is the third time that this has happened. First for me. Luckily the SQL INSERT statement was still in the command buffer.

Syndicate content