Horton

Re: Re: [COMMITTERS] pgsql: Basic Recovery Control functions for use in Hot Standby. Pause,

Heikki Linnakangas <heikki.linnakangas@enterprisedb.com>

View thread
From: Heikki Linnakangas <heikki.linnakangas@enterprisedb.com>
To: Robert Haas <robertmhaas@gmail.com>
Cc: Fujii Masao <masao.fujii@gmail.com>, Tom Lane <tgl@sss.pgh.pa.us>, Magnus Hagander <magnus@hagander.net>, Simon Riggs <simon@2ndquadrant.com>, PostgreSQL-development <pgsql-hackers@postgresql.org>
Date: 2011-03-18T12:27:15Z
Lists: pgsql-hackers

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. Basic Recovery Control functions for use in Hot Standby. Pause, Resume,

On 18.03.2011 14:14, Robert Haas wrote:
> On Fri, Mar 18, 2011 at 3:22 AM, Heikki Linnakangas
> <heikki.linnakangas@enterprisedb.com>  wrote:
>> If recovery target is set to before its consistent, ie. before
>> minRecoveryPoint, we should throw an error before recovery even starts. I'm
>> not sure if we check that at the moment.
>
> I don't see how you could check that anyway.  How do you know where
> you're going to see the given XID/timestamp/named restore point until
> you actually get there?

Oh, good point. I was thinking that the recovery target is a particular 
LSN, but clearly it's not.

>> Not sure what to to do recovery target is beyond minRecoveryPoint and
>> pause_at_recovery_target=true, but the server hasn't been opened for hot
>> standby yet (because it hasn't seen a running-xacts record yet). I agree
>> it's pretty useless and annoying to stop there.
>
> I think the reasonable options are "enter normal running" and "shut down".
>
> In any event, it sounds like someone needs to fix this, and I don't
> know enough to do it.  Can you or Fujii Masao do it?

You could also argue for "log a warning, continue until we can open for 
Hot standby, then pause".

I can write the patch once we know what we want. All of those options 
sound reasonable to me. This is such a corner-case that it doesn't make 
sense to make it user-configurable, though.

-- 
   Heikki Linnakangas
   EnterpriseDB   http://www.enterprisedb.com