Anyone Have a Good PostgreSQL Point in Time Recovery Story?

So I'm writing a talk on PostgreSQL backups and I'm not experienced in PostgreSQL PITR but I need to cover it.

My original slides dissed it and basically gave the reasons not to use it. Feedback from the org head stated I needed to tone that down and demo it.

I'm looking for anecdotal evidence either way. Do you have a PostgreSQL PITR story? I'd love to hear it. Keep it short because short stories are easier to type.

Thank you!

1 comment

[ 2.9 ms ] story [ 12.9 ms ] thread

craigkerstiens 9y ago ↗

What we did at Heroku as well as at Citus is rolled our own using the WAL logs.

We run WAL-E for all clusters and for a PITR we have the ability to restore a base backup then replay the WAL to a specific point. If this happens extremely rarely then doing this manually might absolutely work for you. For a case where it happens more frequently putting in more automation around it can absolutely make sense.