[Bug] [Zeta] Java process does not receive SIGTERM when running SeaTunnel inside container

[qiaoborui]

Search before asking

• I had searched in the issues and found no similar issues.

What happened

When running SeaTunnel inside a Docker/Kubernetes container, the Java process cannot gracefully shut down because it never receives the SIGTERM signal sent by the container runtime.
Currently, the startup script (bin/seatunnel-cluster.sh) launches Java like this (when not using -d):

java ${JAVA_OPTS} -cp ${CLASS_PATH} ${APP_MAIN} ${args}

In this case:
• The shell script itself becomes PID 1 in the container.
• The Java process runs as a child process of the shell.
• When the container is stopped, SIGTERM is only delivered to the shell script (PID 1), not to the Java process.
• The script exits immediately without forwarding the signal, leaving Java as an orphan process.
• Finally, the Java process is killed by SIGKILL after the grace period, preventing SeaTunnel from shutting down gracefully.

This causes issues such as:
• Shutdown hooks in Java never being executed.
• Potential data loss or corruption when SeaTunnel tasks are abruptly killed.

SeaTunnel Version

2.3.11

SeaTunnel Config

none

Running Command

./bin/seatunnel-cluster -r xxx

Error Exception

none

Zeta or Flink or Spark Version

No response

Java or Scala Version

No response

Screenshots

No response

Are you willing to submit PR?

• Yes I am willing to submit a PR!

Code of Conduct

• I agree to follow this project's Code of Conduct

[qiaoborui]

Could you please assign this issue to me? I believe it can be resolved solely at the Bash script level. @Hisoka-X