Greetings,
Please don't top-post on these lists.
* Kirill Reshke (reshkekirill@gmail.com) wrote:
> We have physical backups and we can PITR. But restoring a cluster to some
> point in the past is a bit of a different task: we need our client's
> approval for these operations, since we are a Managed DBs Cloud Provider.
> Will try to ask someone.
Do you have page-level checksums enabled for these PG instances?
Are you able to see if these clusters which are reporting the corruption
have been restored in the past from a backup? What are you using to
perform your backups and perform your restores?
Are you able to see if these clusters have ever crashed and come back up
after by doing WAL replay?
Where I'm heading with these questions is essentially: I suspect either
your backup/restore procedure is broken or you're running on a system
that doesn't properly fsync data. Or possibly both.
Oh, and you should probably have checksums enabled.
Thanks,
Stephen
