With the wrong settings (eg, 1 site backed up every 5 minutes or so), a single site that fails to backup will cause the backup queue to become useless, as the same site will be processed (and fail) over and over again.

A common reason for a site to fail is if it failed to install. That should be adressed as well but sometimes isn't (eg for dev sites). The failed backup should also be addressed but it might not be noticed.

Proposed solutions:

  • Short term (current instances): Increase the backup queue to 5 backups every 5 minutes (or basically more than 1 backup every X minutes). We still get the same failing task every 5 minutes but at least the other backups will run.
  • Long term (code): Maybe try to at least alternate the backup by checking if it was the last backup to ever fail.
  • Other long-term solution: Increase the default backup queue to 5 (or >1), as Aegir's queue system should be able to handle this.

The root problem is always that the backup fails, but this seems like a problem that deserves mitigating.

It could also suffice to send alerts, maybe we could create an alert module if it doesn't already exist (couldn't find one). There's no real problem if you notice the failed backup, the danger is mostly in not getting daily backups for weeks.

I think this problem also applies to the D7 version.

Comments

gboudrias’s picture

gboudrias commented

After writing this I'm thinking an alert module might be the way to go, as increasing the number of backups won't help if they all fail for system reasons.

gboudrias’s picture

gboudrias commented
Version: 6.x-2.x-dev » 7.x-3.x-dev
Priority: Major » Minor

I've found https://www.drupal.org/project/aegir_rules. It appears incomplete but probably a good starting point for alerts.

As far as I'm concerned this is low priority seeing as we can address the root problem. It could be a good idea to test on Aegir3, figure something out and backport it if necessary. At this point I don't think too many people would even see a D6 release.