Horton

Re: Incorrect errno used with %m for backend code

Andres Freund <andres@anarazel.de>

View thread View on postgr.es
From: Andres Freund <andres@anarazel.de>
To: Alvaro Herrera <alvherre@2ndquadrant.com>
Cc: Michael Paquier <michael@paquier.xyz>, Postgres hackers <pgsql-hackers@postgresql.org>
Date: 2018-06-22T19:14:06Z
Lists: pgsql-hackers 
Hi,

On 2018-06-22 11:15:53 -0400, Alvaro Herrera wrote:
> I don't understand the logic in RestoreSlotFromDisk that fsync the file
> prior to reading it.  What are we protecting against?

we could previously have crashed before fsyncing. But we can only rely
on slot contents being durable if we fsync them. Therefore we fsync
before reading, after a crash.


> Anyway, while I think it would be nice to have CloseTransientFile
> restore the original errno if there was one and close goes well, it
> probably unduly complicates its API contract.

Yea, agreed.

Greetings,

Andres Freund

Commits

  1. Address set of issues with errno handling