Thread: Logical decoding on standby

Logical decoding on standby

From
Andreas Joseph Krogh
Date:
Anybody knows if $subject will make it into v11?
 
--
Andreas Joseph Krogh

Re: Logical decoding on standby

From
Andreas Kretschmer
Date:
On 12 March 2018 21:18:22 CET, Andreas Joseph Krogh <andreas@visena.com> wrote:
>Anybody knows if $subject will make it into v11?
> 
>--
>Andreas Joseph Krogh

Why do you think you needs this?

Regards, Andreas


--
2ndQuadrant - The PostgreSQL Support Company


Re: Logical decoding on standby

From
Andreas Kretschmer
Date:
On 13 March 2018 00:58:27 CET, Andreas Kretschmer <andreas@a-kretschmer.de> wrote:
>On 12 March 2018 21:18:22 CET, Andreas Joseph Krogh
><andreas@visena.com> wrote:
>>Anybody knows if $subject will make it into v11?
>> 
>>--
>>Andreas Joseph Krogh
>
>Why do you think you needs this?
>
>Regards, Andreas

Let me explain my question. One of the key aspects of logical replication is, that you can define what to replicate.
Thatwouldn't work in this way, that's why i'm asking. 

Regards, Andreas


--
2ndQuadrant - The PostgreSQL Support Company


Re: Logical decoding on standby

From
"David G. Johnston"
Date:
On Mon, Mar 12, 2018 at 5:08 PM, Andreas Kretschmer <andreas@a-kretschmer.de> wrote:
Let me explain my question. One of the key aspects of logical replication is, that you can define what to replicate. That wouldn't work in this way, that's why i'm asking.

One of the key aspects of "standby" is that it is ready to be a drop-in replacement for the "active" server, and today our logical replication facility is not capable of ensuring that property and won't be for v11 that I can tell.  I'm not certain that said capability is even a goal at present.

IOW, why do you need a "standby" that isn't a drop-in replacement for a primary - i.e., can have a filter on what replicated data it accepts?

David J.

Sv: Re: Logical decoding on standby

From
Andreas Joseph Krogh
Date:
På tirsdag 13. mars 2018 kl. 01:08:03, skrev Andreas Kretschmer <andreas@a-kretschmer.de>:
On 13 March 2018 00:58:27 CET, Andreas Kretschmer <andreas@a-kretschmer.de> wrote:
>On 12 March 2018 21:18:22 CET, Andreas Joseph Krogh
><andreas@visena.com> wrote:
>>Anybody knows if $subject will make it into v11?
>> 
>>--
>>Andreas Joseph Krogh
>
>Why do you think you needs this?
>
>Regards, Andreas

Let me explain my question. One of the key aspects of logical replication is, that you can define what to replicate. That wouldn't work in this way, that's why i'm asking.

Regards, Andreas
 
I have streaming-replication of a cluster containing many databases to a standby-server. I need a reporting-server which only needs a subset of some (large) tables of one database and am planning to use the built-in logical replication for that. It is my understanding that logical replication will cause more wal-traffic so I'm trying to offload wal-traffic from the primary. I thought using logical replication from the standby would help with that but realized it's not supported in v10. Im I wrong in planning this way?
 
--
Andreas Joseph Krogh
CTO / Partner - Visena AS
Mobile: +47 909 56 963
 
Attachment

Re: Logical decoding on standby

From
Andres Freund
Date:
Hi,

On 2018-03-13 01:08:03 +0100, Andreas Kretschmer wrote:
> On 13 March 2018 00:58:27 CET, Andreas Kretschmer <andreas@a-kretschmer.de> wrote:
> >On 12 March 2018 21:18:22 CET, Andreas Joseph Krogh
> ><andreas@visena.com> wrote:
> >>Anybody knows if $subject will make it into v11?

I can't see it happening for v11, there's not even a CF entry in the
last CF.


> >Why do you think you needs this? 
> >
> >Regards, Andreas
> 
> Let me explain my question. One of the key aspects of logical replication is, that you can define what to replicate.
Thatwouldn't work in this way, that's why i'm asking.
 

The subject said logical decoding, not replication. There's a lot of
change data capture type workloads where decoding from the standby is
quite useful. And the design definitely would work for that, we've
explicitly took that into consideration.

Greetings,

Andres Freund


Re: Logical decoding on standby

From
Andreas Kretschmer
Date:

Am 13.03.2018 um 02:40 schrieb Andres Freund:
>
> The subject said logical decoding, not replication. There's a lot of
> change data capture type workloads where decoding from the standby is
> quite useful. And the design definitely would work for that, we've
> explicitly took that into consideration.
>
> Greetings,
>
> Andres Freund
>

Ah, thx, didn't know that.


Regards, Andreas

-- 
2ndQuadrant - The PostgreSQL Support Company.
www.2ndQuadrant.com