Horton

Re: Checkpoint not retrying failed fsync?

Thomas Munro <thomas.munro@enterprisedb.com>

View thread View on postgr.es
From: Thomas Munro <thomas.munro@enterprisedb.com>
To: Andrew Gierth <andrew@tao11.riddles.org.uk>
Cc: Pg Hackers <pgsql-hackers@postgresql.org>
Date: 2018-04-06T01:11:36Z
Lists: pgsql-hackers

Attachments

On Fri, Apr 6, 2018 at 12:56 PM, Thomas Munro
<thomas.munro@enterprisedb.com> wrote:
> After some testing, here is a better one for review.

One problem I thought of about 8 milliseconds after clicking send is
that bms_union() may fail to allocate memory and then you're hosed.
Here is a new version that uses bms_join() instead, because that can't
fail.

-- 
Thomas Munro
http://www.enterprisedb.com

Commits

  1. Don't forget about failed fsync() requests.