Disaster Recovery for a 125 GB Odoo Production Host

The problem

The client's largest Odoo production host carries the ERP for the whole business. It had daily backups, but no tested answer to the question of what happens if the machine itself is gone, and no second location.

• •This is the single most important machine the client runs. It cannot be experimented on
• •Backups that live with the thing they are backing up are not backups. The copy had to leave the provider entirely
• •A cold archive is not a recovery plan on its own. Restoring a database of this size from scratch under pressure takes longer than the business can wait

How we ran it

1. 01Daily database and filestore backups pushed to object storage on separate infrastructure, so losing the provider does not lose the data
2. 02A warm standby kept on a different provider in a different country, so the recovery path does not run through the same failure
3. 03Monitoring on both, because a standby nobody checks is a standby that is quietly broken when it is finally needed
4. 04Treated the production host as read-only for experiments. Anything that needed proving was proven elsewhere first

Ongoing managed service

RAM production host under management

125 GBRAM production host under management

off-site backup cadence

Dailyoff-site backup cadence

providers, in 2 countries

2providers, in 2 countries