Re: Replication server timeout patch

On Wed, Mar 23, 2011 at 6:33 AM, Heikki Linnakangas
<heikki(dot)linnakangas(at)enterprisedb(dot)com> wrote:
> On 16.03.2011 11:11, Fujii Masao wrote:
>>
>> On Wed, Mar 16, 2011 at 4:49 PM, Fujii Masao<masao(dot)fujii(at)gmail(dot)com>
>>  wrote:
>>>
>>> Agreed. I'll change the patch.
>>
>> Done. I attached the updated patch.
>
> I don't much like the API for this. Walsender shouldn't need to know about
> the details of the FE/BE protocol, pq_putbytes_if_available() seems too low
> level to be useful.
>
> I think a better API would be to have a non-blocking version of
> pq_putmessage(). We can make the output buffer in pqcomm.c resizeable, so
> that when the message doesn't fit in the output buffer in pq_putmessage(),
> the buffer is enlarged instead of trying to flush it.
>
> Attached is a patch using that approach. This is a much smaller patch, and
> easier to understand. I'm not totally happy with the walsender main loop, it
> seems to work as it is, but the logic has become quite complicated. Ideas
> welcome on how to simplify that.

Heikki, are you planning to commit this, either with or without
further revisions?

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company