Thread: pgsql: Reset master xmin when hot_standby_feedback disabled.
Reset master xmin when hot_standby_feedback disabled. If walsender has xmin of standby then ensure we reset the value to 0 when we change from hot_standby_feedback=on to hot_standby_feedback=off. Branch ------ master Details ------- http://git.postgresql.org/pg/commitdiff/bd56e74127dea4102d1fc761d65fefbb32146713 Modified Files -------------- doc/src/sgml/protocol.sgml | 6 +++- src/backend/replication/walreceiver.c | 42 ++++++++++++++++++++++++-------- src/backend/replication/walsender.c | 5 +++- 3 files changed, 39 insertions(+), 14 deletions(-)
Hi, On 2013-02-04 10:35:31 +0000, Simon Riggs wrote: > Reset master xmin when hot_standby_feedback disabled. > If walsender has xmin of standby then ensure we > reset the value to 0 when we change from hot_standby_feedback=on > to hot_standby_feedback=off. Unless somebody protests I'd like to backpatch this change to 9.2 and 9.1. This imo is a bugfix, not a new feature. What currently happens < 9.3 when hot_standby_feedback is disabled after being enabled is that the walsender's xmin horizon doesn't increase until disconnected. That's pretty hard to diagnose and can cause severe problems. Greetings, Andres Freund -- Andres Freund http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Training & Services
On 28 May 2014 23:36, Andres Freund <andres@2ndquadrant.com> wrote: > On 2013-02-04 10:35:31 +0000, Simon Riggs wrote: >> Reset master xmin when hot_standby_feedback disabled. >> If walsender has xmin of standby then ensure we >> reset the value to 0 when we change from hot_standby_feedback=on >> to hot_standby_feedback=off. > > Unless somebody protests I'd like to backpatch this change to 9.2 and > 9.1. This imo is a bugfix, not a new feature. What currently happens < > 9.3 when hot_standby_feedback is disabled after being enabled is that > the walsender's xmin horizon doesn't increase until disconnected. That's > pretty hard to diagnose and can cause severe problems. Good thought. I'll backpatch. -- Simon Riggs http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Training & Services
On 2014-05-28 23:39:56 +0100, Simon Riggs wrote: > On 28 May 2014 23:36, Andres Freund <andres@2ndquadrant.com> wrote: > > > On 2013-02-04 10:35:31 +0000, Simon Riggs wrote: > >> Reset master xmin when hot_standby_feedback disabled. > >> If walsender has xmin of standby then ensure we > >> reset the value to 0 when we change from hot_standby_feedback=on > >> to hot_standby_feedback=off. > > > > Unless somebody protests I'd like to backpatch this change to 9.2 and > > 9.1. This imo is a bugfix, not a new feature. What currently happens < > > 9.3 when hot_standby_feedback is disabled after being enabled is that > > the walsender's xmin horizon doesn't increase until disconnected. That's > > pretty hard to diagnose and can cause severe problems. > > Good thought. I'll backpatch. Cool. Happy to escape my first backpatch for a little while longer :)! Greetings, Andres Freund -- Andres Freund http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Training & Services
Andres Freund wrote: > On 2014-05-28 23:39:56 +0100, Simon Riggs wrote: > > Good thought. I'll backpatch. > > Cool. Happy to escape my first backpatch for a little while longer :)! Meh. Pretty sure it'll be a no-brainer for you. -- Álvaro Herrera http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Training & Services