Horton

Re: Spurious "apparent wraparound" via SimpleLruTruncate() rounding

Noah Misch <noah@leadboat.com>

View thread View on postgr.es
From: Noah Misch <noah@leadboat.com>
To: pgsql-hackers@postgresql.org
Date: 2019-02-11T05:13:49Z
Lists: pgsql-hackers 
On Sat, Feb 02, 2019 at 03:38:22AM -0500, Noah Misch wrote:
> The main consequence is the false alarm.  A prudent DBA will want to react to
> true wraparound, but no such wraparound has occurred.  Also, we temporarily
> waste disk space in pg_xact.  This feels like a recipe for future bugs.  The
> fix I have in mind, attached, is to change instances of
> ctl->PagePrecedes(FIRST_PAGE_OF_SEGMENT, ROUNDED_cutoffPage) to
> ctl->PagePrecedes(LAST_PAGE_OF_SEGMENT, cutoffPage).  I'm inclined not to
> back-patch this; does anyone favor back-patching?

To avoid wasting more of anyone's time: that patch is bad; I'll update this
thread when I have something better.

Commits

  1. Prevent excess SimpleLruTruncate() deletion.

  2. Fix unlinking of SLRU segments.

  3. Defer flushing of SLRU files.

  4. Change XID and mxact limits to warn at 40M and stop at 3M.