On Wed, Aug 30, 2017 at 6:45 AM, Andres Freund <andres@anarazel.de> wrote:
On 2017-08-30 12:52:26 +0900, Michael Paquier wrote: > On Wed, Aug 30, 2017 at 11:20 AM, Peter Eisentraut > <peter.eisentraut@2ndquadrant.com> wrote: > > On 8/29/17 20:36, Andres Freund wrote: > >> So the question is whether we want {max,min}_wal_size be sized in > >> multiples of segment sizes or as a proper byte size. I'm leaning > >> towards the latter. > > > > I'm not sure what the question is or what its impact would be. > > FWIW, I get the question as: do we want the in-memory values of > min/max_wal_size to be calculated in MB (which is now the case) or > just bytes. Andres tends for using bytes.
Not quite. There's essentially two things:
1) Currently the default for {min,max}_wal_size depends on the segment size. Given that the segment size is about to be configurable, that seems confusing. 2) Currently wal_segment_size is measured in GUC_UNIT_XBLOCKS, which requires us to keep two copies of the underlying variable, one in XBLOCKS one in bytes. I'd rather just have the byte variant.
I'd say we definitely want the "user interface" to be in bytes(/mbytes/gbytes etc). We used to have that in segments and it was quite confusing for a lot of new uers, and seemed very silly...
Also agreed that (1) above seems very confusing. Going to using bytes all the way seems a lot more clear.