Re: auto-sizing wal_buffers

On Thu, Jan 13, 2011 at 6:02 PM, Josh Berkus  wrote:
>
>> Depends what people want to do.  We could make the default "0kB", and
>> define that to mean "auto-tune", or we could remove the parameter
>> altogether.  I think I was envisioning the latter, but if people are
>> hesitant to do that we could do the former instead.
>
> Unfortunately, we might still need a manual parameter for override
> because of the interaction between wal_buffers and
> synchronous_commit=off, since it sets the max size of the unflushed data
> buffer.  Discuss?

Do we have any evidence there's actually a problem in that case, or
that a larger value of wal_buffers solves it?  I mean, the background
writer is going to start a background flush as quickly as it can...

> And the "auto" setting should be -1, not 0kB.  We use -1 for "use
> default" for several other GUCs.

No can do.  Gotta have things in the same units.

> Other than that, I think Greg's numbers are fine, and strongly support
> having one less thing to tune.

OK.

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company