Re: Standbys, txid_current_snapshot, wraparound

Marko Kreen <markokr@gmail.com>

From: Marko Kreen <markokr@gmail.com>

To: Simon Riggs <simon@2ndQuadrant.com>

Cc: Daniel Farina <daniel@heroku.com>, pgsql-hackers <pgsql-hackers@postgresql.org>, Andrey Velikoredchanin <uncleandyv@gmail.com>

Date: 2012-03-29T14:29:37Z

Lists: pgsql-hackers

On Thu, Mar 29, 2012 at 03:23:01PM +0100, Simon Riggs wrote:
> On Thu, Mar 29, 2012 at 3:04 PM, Marko Kreen <markokr@gmail.com> wrote:
> > Next question: how can flipping archive_mode on and off,
> > with restarts, near wraparound point, break epoch on master?
> >
> >  http://lists.pgfoundry.org/pipermail/skytools-users/2012-March/001609.html
> >
> > Any ideas?
> >
> > Could WAL playback from backend with different archive_mode setting
> > cause it somehow?
> 
> The current code would allow it, if you managed to go for 4 billion
> xids between checkpoints. If that happened, it would give an off by
> one error each time it happened.
> 
> I don't believe that is possible.

Yes, that does not sound like probable case.

-- 
marko