Hold off on AWS Aurora and Postgres until further notice

We've had a sudden crash of the underlying storage (apparently) on one of our production workloads - and because the crash is in the underlying storage the MultiAZ setting makes no difference. Neither the primary nor the replica work.

The point in time restore can't complete a restore to the last available point, so this looks like data loss to me. Points a earlier in time work fine, so might be an issue with a recent patch or failure.

I'm currently in conversation with support, will update when I have more information.