Why You Should Avoid Using kill ‑9 to Stop Programs

kill sends a signal to a process; SIGKILL (signal 9) forces immediate termination without cleanup. Developers often use kill -9 <pid> to kill stubborn processes, assuming it is harmless.

Problems Caused by kill ‑9

In production environments, force‑killing a process can lead to fatal errors that are hard to diagnose because the program does not have a chance to release resources, flush buffers, or update state.

For example, consider a money‑transfer operation where account A deducts funds and account B receives them. If the process handling the transfer is killed abruptly (or the system loses power), the transaction may be left half‑complete. With the InnoDB engine, which provides transactional guarantees, the operation would roll back, preserving consistency. However, with the MyISAM engine, which lacks transaction support, the debit from A may be recorded while the credit to B is lost, resulting in a bad debt—a disaster in production.

This abrupt termination is analogous to a sudden power outage; kill ‑9 effectively cuts power to the process, preventing any graceful shutdown logic from running.

Because of these risks, many production teams prohibit the use of kill -9 <pid> for cleaning up processes, preferring gentler signals (e.g., SIGTERM) that allow the program to perform necessary cleanup and maintain data integrity.

The article concludes by urging developers to avoid kill ‑9 in production and to discuss safer termination practices.