Posted by defiantbyte on July 26, 2013 at 11:09am
I recently tried moving a site backup from one server to the other but ran into an issue with the database.sql file. It looks like it doesn't have all of the tables in it. The last table in the dump is 'cache_form' and there is no sign of 'system', 'node', etc..
Has anyone seen something like this before?
Comments
Some more information
This is actually only happening with a single octopus instance. Another instance on the same server is outputting the full database backup.
An item of note with the problem instance is they store very large EXE files within the site files. I'm wondering if there is a memory / backup size limit?
This happens often - never trust a BOA backup
If your server isn't up to the task - if the cpu load is too high, or if there isn't enough RAM available when the backup is being taken, then it won't give you a complete backup - it will timeout before the backup is completed, or there will be some other problem with the backup. I have seen this over and over for 2 years now. So, you need to occasionally log into the site, create a db backup and make an ftp backup of the files you need to recreate the site on another server, etc ...
cache_form bin is the culprit
This is nothing related to resources on the machine/vps.
You may want to chime in here: https://drupal.org/node/2088073
BOA includes this more aggressive cache clear action in head: https://github.com/omega8cc/provision/commit/3faba67adce970653862f05560d...
Also, you should remember
Also, you should remember that BOA does purge all cache bins when it is doing nightly backup for all databases, and they are reliably archived in the /data/disk/arch/sql directory.