Kevin Grittner wrote:
> Bruce Momjian <bruce@momjian.us> wrote:
>
> > I assume we send a full 8k to the controller, and a failure during
> > that write is not registered as a write.
>
> On what do you base that assumption? I assume that we send a full
> 8K to the OS cache, and the file system writes disk sectors
> according to its own algorithm. With either platters or BBU cache,
> the data is persisted on fsync; why do you see a risk with one but
> not the other?
Now that is an interesting question. We write 8k to the kernel, but the
kernel doesn't have to honor those write sizes, so while we probably
can't get a partial 512-byte block written to disk with an BBU (that
isn't cleanup up by the BBU on reboot), we could get some 512-byte
blocks of an 8k written and others not.
I agree you are right and a BBU does not mean you can safely turn off
full_page_writes.
--
Bruce Momjian <bruce@momjian.us> http://momjian.us
EnterpriseDB http://enterprisedb.com
+ It's impossible for everything to be true. +