Re: bug in SignalSomeChildren

On Thu, Jan 20, 2011 at 9:53 PM, Bernd Helmle <mailings@oopsware.de> wrote:
> I had a look at this for the current CF and the patch looks reasonable to
> me. Some testing shows that the changes are working as intended (at least,
> the wal sender actually receives now signals from SignalSomeChildren() as
> far as the DEBUG4 output shows).

Thanks for the review and test!

> Maybe we should put in a small comment, why
> we special case BACKEND_TYPE_ALL (following Tom's comment about expensive
> shared memory access and IsPostmasterChildWalSender()).

I added the following small comment. Patch attached.

+        /*
+         * Since target == BACKEND_TYPE_ALL is the most common case,
+         * we test it first and avoid touching shared memory for
+         * every child.
+         */

> Question for my understanding:
>
> While reading the small patch, i realized that there's no
> BACKEND_TYPE_WALRECV or similar. If i understand correctly there's no need
> to handle it this way, since there's only one wal receiver process per
> instance?

Yes. But also that's because walreceiver is an auxiliary process (like
bgwriter and
walwriter ..etc) but not a backend.

Regards,

--
Fujii Masao
NIPPON TELEGRAPH AND TELEPHONE CORPORATION
NTT Open Source Software Center