Horton

RE: [Proposal] Add foreign-server health checks infrastructure

Hayato Kuroda (Fujitsu) <kuroda.hayato@fujitsu.com>

View thread View on postgr.es
From: "kuroda.hayato@fujitsu.com" <kuroda.hayato@fujitsu.com>
To: 'Kyotaro Horiguchi' <horikyota.ntt@gmail.com>
Cc: "onderkalaci@gmail.com" <onderkalaci@gmail.com>, "andres@anarazel.de" <andres@anarazel.de>, "masao.fujii@oss.nttdata.com" <masao.fujii@oss.nttdata.com>, "pgsql-hackers@lists.postgresql.org" <pgsql-hackers@lists.postgresql.org>, "Shinya11.Kato@oss.nttdata.com" <Shinya11.Kato@oss.nttdata.com>, "zyu@yugabyte.com" <zyu@yugabyte.com>
Date: 2022-10-17T12:04:07Z
Lists: pgsql-hackers 
Dear Horiguchi-san,

> Might be on slight different direction, but it looks to me a bit too
> much to use WaitEventSet to check only if a socket is live or not.
> 
> A quick search in the tree told me that we could use pqSocketCheck()
> instead, and I think it would be the something that "could potentially
> go into libpq-fe.h" as Önder mentioned, if I understand what he said
> correctly.

Based on your suggestion, I tried to add like following function to fe-misc.c:

```
int
PQconncheck(PGconn *conn)
{
	/* Raise an ERROR if invalid socket has come */
	if (conn == NULL ||
		PQsocket(conn) == PGINVALID_SOCKET)
		return -1;

	return pqSocketCheck(conn, 1, 1, -1);
}
```

... and replace pgfdw_connection_check_internal() to PQconncheck(),
but it did not work well.
To be exact, pqSocketCheck() said the socket was "readable" and "writable"
even if the connection has been killed.

IIUC, pqSocketCheck () calls pqSocketPoll(),
and in the pqSocketPoll() we poll()'d the POLLIN or POLLOUT event.
But according to [1], we must wait POLLRDHUP event,
so we cannot reuse it straightforward.

If we really want to move the checking function anyway,
we must follow AddWaitEventToSet() and some WaitEventAdjust functions.
I'm not sure whether it is really good.

[1]: https://man7.org/linux/man-pages/man2/poll.2.html

Best Regards,
Hayato Kuroda
FUJITSU LIMITED

Commits

  1. postgres_fdw: Extend postgres_fdw_get_connections to return user name.

  2. postgres_fdw: Fix bug in connection status check.

  3. postgres_fdw: Add connection status check to postgres_fdw_get_connections().

  4. postgres_fdw: Add "used_in_xact" column to postgres_fdw_get_connections().

  5. doc: Enhance documentation for postgres_fdw_get_connections() output columns.

  6. Avoid reference to nonexistent array element in ExecInitAgg().

  7. Add WL_SOCKET_CLOSED for socket shutdown events.

  8. postgres_fdw: reestablish new connection if cached one is detected as broken.