Hi,
On 2019-03-06 11:56:26 +0900, Michael Paquier wrote:
> Windows is not concurrent safe by default when it comes to handle
> writes, and Postgres uses a wrapper for open() which makes use of some
> concurrent-safe options when handling files to avoid such problems.
I don't think this has anything to do with the problem. Isn't this the
known problem around truncation that Tom has recently talked about fixing?
[1] https://www.postgresql.org/message-id/2348.1544474335%40sss.pgh.pa.us
Greetings,
Andres Freund