Re: An extra error for client disconnection on Windows

Haribabu Kommi <kommi(dot)haribabu(at)gmail(dot)com> writes:
> On Thu, Jun 2, 2016 at 6:51 PM, Kyotaro HORIGUCHI <
> horiguchi(dot)kyotaro(at)lab(dot)ntt(dot)co(dot)jp> wrote:
>> After a process termination without PQfinish() of a client,
>> server emits the following log message not seen on Linux boxes.
>> LOG: could not receive data from client: An existing connection was
>> forcibly closed by the remote host.
>>
>> This patch translates WSAECONNRESET of WSARecv to an EOF so that
>> pgwin32_recv behaves the same way with Linux.

> Marked the patch as "ready for committer".

Windows is not my platform, but ... is this actually an improvement?
I'm fairly concerned that this change would mask real errors that ought
to get logged. I don't know that that's an okay price to pay for
suppressing a log message when clients violate the protocol.

According to
https://msdn.microsoft.com/en-us/library/windows/desktop/ms740668(v=vs.85).aspx
WSAECONNRESET means:

An existing connection was forcibly closed by the remote host. This
normally results if the peer application on the remote host is
suddenly stopped, the host is rebooted, the host or remote network
interface is disabled, or the remote host uses a hard close (see
setsockopt for more information on the SO_LINGER option on the remote
socket). This error may also result if a connection was broken due to
keep-alive activity detecting a failure while one or more operations
are in progress. Operations that were in progress fail with
WSAENETRESET. Subsequent operations fail with WSAECONNRESET.

(The description of WSAENETRESET, on the same page, indicates that the
last two sentences apply only to the keep-alive failure case.)

So this change would deal nicely with the "peer application on the remote
host is suddenly stopped" case, at the price of being not nice about any
of the other cases. Not convinced it's a good tradeoff.

regards, tom lane