Hi,
On 2019-10-24 16:31:39 -0700, Maciek Sakrejda wrote:
> On Thu, Oct 24, 2019 at 2:25 PM Andres Freund wrote:
> > Note that the buffer access stats do *not* count the number of distinct
> > buffers accessed, but that they purely the number of buffer
> > accesses.
>
> You mean, even within a si
On Thu, Oct 24, 2019 at 2:25 PM Andres Freund wrote:
> Note that the buffer access stats do *not* count the number of distinct
> buffers accessed, but that they purely the number of buffer
> accesses.
You mean, even within a single node? That is, if a node hits a block ten
times, that counts as t
Hi,
On 2019-10-21 23:18:32 -0700, Maciek Sakrejda wrote:
> I ran across an EXPLAIN plan and had some questions about some of its
> details. The BUFFERS docs say
>
> >The number of blocks shown for an upper-level node includes those used by
> all its child nodes.
>
> I initially assumed this woul
Also, I noticed that in this plan, the root (again, an Aggregate) has 0
Temp Read Blocks, but two of its children (two of the ModifyTable nodes)
have non-zero Temp Read Blocks. Again, this contradicts the documentation,
as these costs are stated to be cumulative. Any ideas?
Thanks,
Maciek
Hello,
I ran across an EXPLAIN plan and had some questions about some of its
details. The BUFFERS docs say
>The number of blocks shown for an upper-level node includes those used by
all its child nodes.
I initially assumed this would be cumulative, but I realized it's probably
not because some o
