Hi,
On 2018-02-28 17:14:18 -0500, Peter Eisentraut wrote:
> I can see why you'd want that, but as a DBA, I don't necessarily want
> all of that recorded, especially in a quasi-permanent way.
Huh? You're arguing that we should make it easier for DBAs to hide
potential causes of corruption? I fail to see when that's necessary /
desirable? That a cluster ran with fsync=off isn't an embarassing thing,
nor does it contain private data...
Greetings,
Andres Freund