Re: Standalone synchronous master

Andres Freund <andres@2ndquadrant.com>

From: Andres Freund <andres@2ndquadrant.com>
To: Jim Nasby <jim@nasby.net>
Cc: "Joshua D. Drake" <jd@commandprompt.com>, Hannu Krosing <hannu@2ndQuadrant.com>, Florian Pflug <fgp@phlo.org>, Josh Berkus <josh@agliodbs.com>, Bruce Momjian <bruce@momjian.us>, Amit Kapila <amit.kapila16@gmail.com>, MauMau <maumau307@gmail.com>, Tom Lane <tgl@sss.pgh.pa.us>, Kevin Grittner <kgrittn@ymail.com>, Heikki Linnakangas <hlinnakangas@vmware.com>, Simon Riggs <simon@2ndquadrant.com>, Rajeev rastogi <rajeev.rastogi@huawei.com>, pgsql-hackers <pgsql-hackers@postgresql.org>
Date: 2014-01-13T21:18:21Z
Lists: pgsql-hackers
On 2014-01-13 15:14:21 -0600, Jim Nasby wrote:
> On 1/13/14, 12:21 PM, Joshua D. Drake wrote:
> >
> >On 01/13/2014 10:12 AM, Hannu Krosing wrote:
> >>>>In other words, if we're going to have auto-degrade, the most
> >>>>intelligent place for it is in
> >>>>RepMgr/HandyRep/OmniPITR/pgPoolII/whatever.  It's also the *easiest*
> >>>>place.  Anything we do *inside* Postgres is going to have a really,
> >>>>really hard time determining when to degrade.
> >>>+1
> >>>
> >>>This is also how 2PC works, btw - the database provides the building
> >>>blocks, i.e. PREPARE and COMMIT, and leaves it to a transaction manager
> >>>to deal with issues that require a whole-cluster perspective.
> >>>
> >>
> >>++1
> >
> >+1
> 
> Josh, what do you think of the upthread idea of being able to recover in-progress transactions that are waiting when we turn off sync rep? I'm thinking that would be a very good feature to have... and it's not something you can easily do externally.

I think it'd be a fairly simple patch to re-check the state of syncrep
config in SyncRepWaitForLsn(). Alternatively you can just write code to
iterate over the procarray and sets Proc->syncRepState to
SYNC_REP_WAIT_CANCELLED or such.

Greetings,

Andres Freund

-- 
 Andres Freund	                   http://www.2ndQuadrant.com/
 PostgreSQL Development, 24x7 Support, Training & Services