高健 escribió:
> ...
> Checkpoints are fairly expensive, first because they require writing out
> all currently dirty buffers, and second because they result in extra
> subsequent WAL traffic as discussed above.
> ...
>
> What confused me is that: (checkpoint)result in extra subsequent WAL
> traffic as discussed above...
>
> I haven't found any more information can describe it in the "above" of that
> page.
It means that the first change following the checkpoint that affects any
particular page will require a full page image of that page to be
written to WAL. See the discussion in the preceding paragraph about
"full_page_writes". It's not the checkpoint itself that writes this
extra WAL, but the rest of the system.
--
Álvaro Herrera http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services
