public inbox for [email protected]  

help / color / mirror / Atom feed

From: jiye <[email protected]>
To: [email protected]
Subject: FDW connection drops with "Connection timed out" during async append query due to TCP receive buffer filling up
Date: Tue, 10 Mar 2026 18:17:44 +0800 (CST)
Message-ID: <[email protected]> (raw)

Hello:


I encountered an issue with foreign table asynchronous scans.


I'm testing on PostgreSQL 17 in a Windows WSL environment, with TCP keepalive settings configured as follows:
configure:
tcp_keepalives_idle = 2
tcp_keepalives_interval = 2
tcp_keepalives_count = 3
tcp_user_timeout = 9


The foreign table was created with:
sql:
CREATE TABLE ftab (a int, data varchar);
I inserted test data using:
sql:
INSERT INTO ftab
SELECT generate_series(1, 10000),
       substring(md5(random()::text) || repeat(md5(random()::text), 127), 1, 40960);
Then I executed an async append query that unions with a simple query:
sql:
SELECT f1(), 'local-data'
UNION ALL
SELECT * FROM f_ftab
LIMIT 2;
Note: f1() is a function that sleeps for 10 seconds, created by:
sql
CREATE OR REPLACE FUNCTION f1()
RETURNS integer
LANGUAGE plpgsql AS $$
BEGIN
    PERFORM pg_sleep(10);
    RETURN 1;
END; $$;
The abnormal behavior was reproduced: the foreign data server connection was disconnected, with the following error logs:
text
2026-03-10 17:52:55.969 CST [820] LOG: statement: FETCH 100 FROM c1
2026-03-10 17:52:56.495 CST [820] LOG: could not receive data from client: Connection timed out
2026-03-10 17:52:56.495 CST [820] LOG: unexpected EOF on client connection with an open transaction
The async foreign table query failed with the error:
text
ERROR: server closed the connection unexpectedly
        This probably means the server terminated abnormally
        before or while processing the request.
invalid socket
CONTEXT: remote SQL command: SELECT a, data FROM public.ftab
WARNING: no connection to the server
This is a minimal working example. In practice, if the local table scan takes too long and the foreign table has sufficiently wide rows, this issue may reproduce.
In my understanding, when performing a local sequential scan, the PostgreSQL backend fetches data from the local plan without fetching any data from the FDW. As a result, the TCP receive buffer may become full, causing the FDW connection to be disconnected.
I believe this is a minor issue. How can I resolve this problem?

reply

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Reply to all the recipients using the --to and --cc options:
  reply via email

  To: [email protected]
  Cc: [email protected], [email protected]
  Subject: Re: FDW connection drops with "Connection timed out" during async append query due to TCP receive buffer filling up
  In-Reply-To: <[email protected]>

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

This inbox is served by agora; see mirroring instructions
for how to clone and mirror all data and code used for this inbox