Re: [bug fix] Produce a crash dump before main() on Windows

On Mon, Feb 26, 2018 at 04:06:48AM +0000, Tsunakawa, Takayuki wrote:
> As for PG11+, I agree that we want to always leave WER on. That is,
> call SetErrorMode(SEM_FAILCRITICALERRORS) but not specify
> SEM_NOGPFAULTERRORBOX. The problem with the current specification of
> PostgreSQL is that the user can only get crash dumps in a fixed folder
> $PGDATA\crashdumps. That location is bad because the crash dumps will
> be backed up together with the database cluster without the user
> noticing it. What's worse, the crash dumps are large. With WER, the
> user can control the location and size of crash dumps.

I have not looked by myself at the problems around WER and such so I
don't have a clear opinion, but including crash dumps within backups is
*bad*. We have a set of filtering rules in basebackup.c and pg_rewind
since v11, we could make use of it and remove all contents from
crashdumps/ from what gets backed up or rewound. excludeDirContents
creates empty folders when those are listed, so we'd want to not create
an empty folder in Windows backups, which makes a bit more more
difficult.
--
Michael