From: Thomas Munro <thomas.mu...@gmail.com>
> I changed my mind. Let's commit the pleasingly simple Linux-only feature for
> now, and extend to it to send some kind of no-op message in a later release.
> So this is the version I'd like to go with.
> Objections?
+1, as some of our users experienced the problem that the server kept
processing (IIRC, a buggy PL/pgSQL procedure that loops indefinitely) after
they killed the client.
TBH, Linux and Windows will be sufficient. But I'm for providing a good
feature on a specific OS first.
(1)
+ rc = poll(&pollfd, 1, 0);
+ if (rc < 0)
+ {
+ elog(COMMERROR, "could not poll socket: %m");
+ return false;
I think it's customary to use ereport() and errcode_for_socket_access().
(2)
pq_check_connection()
Following PostmasterIsAlive(), maybe it's better to name it as
pq_connection_is_alive() pq_client_is_alive(), or pq_frontend_is_alive() like
the pqcomm.c's head comment uses the word frontend?
(3)
#include <limits.h>
+#include <poll.h>
#ifndef WIN32
#include <sys/mman.h>
#endif
poll.h should be included between #ifndef WIN32 and #endif?
(4)
I think the new GUC works for walsender as well. If so, how do we explain the
relationship between the new GUC and wal_receiver_timeout and recommend the
settings of them?
Regards
Takayuki Tsunakawa
