[HACKERS] segment size depending *_wal_size defaults (was increasing thedefault WAL segment size)

Hi,

intentionally breaking the thread here, I want this one point to get
a bit wider audience.

The excerpt of the relevant discussion is:

On 2017-08-23 12:13:15 +0530, Beena Emerson wrote:
> >> +     /* set default max_wal_size and min_wal_size */
> >> +     snprintf(repltok, sizeof(repltok), "min_wal_size = %s",
> >> +                      pretty_wal_size(DEFAULT_MIN_WAL_SEGS));
> >> +     conflines = replace_token(conflines, "#min_wal_size = 80MB", repltok);
> >> +
> >> +     snprintf(repltok, sizeof(repltok), "max_wal_size = %s",
> >> +                      pretty_wal_size(DEFAULT_MAX_WAL_SEGS));
> >> +     conflines = replace_token(conflines, "#max_wal_size = 1GB", repltok);
> >> +
> >
> > Hm. So postgresql.conf.sample values are now going to contain misleading
> > information for clusters with non-default segment sizes.
> >
> > Have we discussed instead defaulting min_wal_size/max_wal_size to a
> > constant amount of megabytes and rounding up when it doesn't work for
> > a particular segment size?
> 
> This was not discussed.
> 
> In the original code, the min_wal_size and max_wal_size are computed
> in the guc.c for any wal_segment_size set at configure.
> 
>     {
>         {"min_wal_size", PGC_SIGHUP, WAL_CHECKPOINTS,
>             gettext_noop("Sets the minimum size to shrink the WAL to."),
>             NULL,
>             GUC_UNIT_MB
>         },
>         &min_wal_size_mb,
>         5 * (XLOG_SEG_SIZE / (1024 * 1024)), 2, MAX_KILOBYTES,
>         NULL, NULL, NULL
>     },
> 
>     {
>         {"max_wal_size", PGC_SIGHUP, WAL_CHECKPOINTS,
>             gettext_noop("Sets the WAL size that triggers a checkpoint."),
>             NULL,
>             GUC_UNIT_MB
>         },
>         &max_wal_size_mb,
>         64 * (XLOG_SEG_SIZE / (1024 * 1024)), 2, MAX_KILOBYTES,
>         NULL, assign_max_wal_size, NULL
>     },
> 
> Hence I have retained the same calculation for min_wal_size and
> max_wal_size. If we get consensus for fixing a default and updating
> when required, then I will change the code accordingly.

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.

- Andres