On 12/01/16 15:11, Russell Coker via luv-main wrote:
When running at VPAC the watchdog table in the Drupal database took up 20G of disk space and as it was in the ibdata1 file there was no good option for freeing the space even after deleting most rows. As part of the migration to the new server I dumped all MySQL databases and imported them. That solved the disk space problem and as I now have MySQL configured for a file per InnoDB table I will always be able to OPTIMIZE TABLE if I have that happen again.
echo "delete from watchdog where timestamp < unix_timestamp() - 604800;" | mysql luv_drupal
The above command is in a cron job to delete the old entries from the watchdog table. It would be good if we could fix the Drupal problem that's causing those entries (but I don't know how to do it or have time to learn). It would also be good if we could configure Drupal to not store millions of rows in that table (it's supposed to automatically delete old rows but doesn't). Assistance from Drupal experts is welcome.
Drupal's watchdog table is the default. It's friendly for small sites and shared hosting but otherwise is likely to be a poor choice. You're better off logging to syslog, and configuring syslog to direct those entries into a separate log file, with appropriate logrotate rules. https://www.drupal.org/documentation/modules/syslog If the watchdog table is used, you should have a cron job to clear out old entries. Its size shouldn't get to megabytes, let alone gigabytes. I think this is part of the default arrangements for drupal. Is your drupal cron running at all? Regards, Andrew McNaughton