Re: Sync Rep Design

On Dec 30, 2010, at 3:27 PM, Robert Haas wrote:
>> synchronous_replication (boolean)
>> Specifies whether transaction commit will wait for WAL records
>> to be replicated before the command returns a "success"
>> indication to the client.
>
> The word "replicated" here could be taken to mean different things,
> most obviously:
>
> - slave has received the WAL
> - slave has fsync'd the WAL
> - slave has applied the WAL

I think that comment is valid for the entire set of docs, actually. The document goes out of its way to avoid simple phrases like "replicated", but doesn't spell out exactly what is happening, ie:

"Synchronous replication offers the ability to guarantee that all changes
made by a transaction have been transferred to at least one remote
standby server. This is an extension to the standard level of durability
offered by a transaction commit. This is referred to as semi-synchronous
replication."

Reading that, I'm left with the sense that this isn't a simple matter of "Oh, the data has been replicated to the slave before commit returns", but nothing does a good job of clearly explaining what the distinction is and what it means. This section:

"The guarantee we offer is that the application will not receive
explicit acknowledgement of the successful commit of a transaction until
the WAL data is known to be safely received by the standby. Hence this
mechanism is technically "semi synchronous" rather than "fully
synchronous" replication."

does provide some enlightenment, but it's at the end of the section. I think it would be best if there was a section right at the beginning that talked about the data quality issue of sync replication and how we're avoiding it with our semi-sync solution.
--
Jim C. Nasby, Database Architect jim(at)nasby(dot)net
512.569.9461 (cell) http://jim.nasby.net