[FLINK-33251] SQL Client query execution aborts after a few seconds: ConnectTimeoutException

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 1.18.0, 1.17.1
• Component/s: Table SQL / Client

Environment

Macbook Pro

Apple M1 Max

$ uname -a
Darwin asgard08 23.0.0 Darwin Kernel Version 23.0.0: Fri Sep 15 14:41:43 PDT 2023; root:xnu-10002.1.13~1/RELEASE_ARM64_T6000 arm64

$ java --version
openjdk 11.0.20.1 2023-08-24
OpenJDK Runtime Environment Homebrew (build 11.0.20.1+0)
OpenJDK 64-Bit Server VM Homebrew (build 11.0.20.1+0, mixed mode)
$ mvn --version
Apache Maven 3.9.5 (57804ffe001d7215b5e7bcb531cf83df38f93546)
Maven home: /opt/homebrew/Cellar/maven/3.9.5/libexec
Java version: 11.0.20.1, vendor: Homebrew, runtime: /opt/homebrew/Cellar/openjdk@11/11.0.20.1/libexec/openjdk.jdk/Contents/Home
Default locale: en_GB, platform encoding: UTF-8
OS name: "mac os x", version: "14.0", arch: "aarch64", family: "mac"

• Assignee: Unassigned
• Reporter: Robin Moffatt
• Created: 12/Oct/23 09:56
• Updated: 28/May/24 10:27

Description

If I run a streaming query from an unbounded connector from the SQL Client, it bombs out after ~15 seconds.

[ERROR] Could not execute SQL statement. Reason:
org.apache.flink.shaded.netty4.io.netty.channel.ConnectTimeoutException: connection timed out: localhost/127.0.0.1:52596

This doesn't happen on 1.16.2. It does happen on 1.17.1 and 1.18 that I have just built locally (git repo hash `9b837727b6d`).

The corresponding task's status in the Web UI shows as `CANCELED`.

—

To reproduce

Launch local cluster and SQL client

➜  flink-1.18-SNAPSHOT ./bin/start-cluster.sh 
Starting cluster.
Starting standalonesession daemon on host asgard08.
Starting taskexecutor daemon on host asgard08.
➜  flink-1.18-SNAPSHOT ./bin/sql-client.sh
[…]
Flink SQL>

Set streaming mode and result mode

Flink SQL> SET 'execution.runtime-mode' = 'STREAMING';
[INFO] Execute statement succeed.

Flink SQL> SET 'sql-client.execution.result-mode' = 'changelog';
[INFO] Execute statement succeed.

Define a table to read data from CSV files in a folder

CREATE TABLE firewall (
  event_time STRING,
  source_ip  STRING,
  dest_ip    STRING,
  source_prt INT,
  dest_prt   INT
) WITH (
  'connector' = 'filesystem',
  'path' = 'file:///tmp/firewall/',
  'format' = 'csv',
  'source.monitor-interval' = '1' -- unclear from the docs what the unit is here
);

Create a CSV file to read in

$ mkdir /tmp/firewall

$ cat > /tmp/firewall/data.csv <<EOF
2018-05-11 00:19:34,151.35.34.162,125.26.20.222,2014,68
2018-05-11 22:20:43,114.24.126.190,21.68.21.69,379,1619
EOF

Run a streaming query

SELECT * FROM firewall;

You will get results showing (and if you add another data file it will show up) - but after ~30 seconds the query aborts and throws an error back to the user at the SQL Client prompt

[ERROR] Could not execute SQL statement. Reason:
org.apache.flink.shaded.netty4.io.netty.channel.ConnectTimeoutException: connection timed out: localhost/127.0.0.1:58470

Flink SQL>