Learn how to notify users, send polite signals, and escalate to forced termination without risking data loss. 18.11.2025 | reading time: 2 min Stuck or idle user sessions block maintenance and sometimes corrupt data; this guide shows commands and a safe escalation path so an admin can notify users first, send a polite termination, and force-kill only when necessary. Practical case: orphaned SSH session An admin finds an old SSH session from "alice" that prevents unmounting a home share; first notify, then terminate politely, then force if needed, for example: ```sh $ who alice pts/1 2025-11-18 09:12 (192.0.2.10) bob pts/2 2025-11-18 09:30 (192.0.2.11) $ wall "Server maintenance in 5 minutes. Save work." Broadcast message from root@host (pts/0) at 10:00 ... Server maintenance in 5 minutes. Save work. $ pkill -TERM -u alice $ sleep 3 $ who bob pts/2 2025-11-18 09:30 (192.0.2.11) ``` Signals, sessions and edge cases Start with SIGTERM to let processes exit cleanly and only use SIGKILL when TERM fails because SIGKILL prevents cleanup; watch for processes stuck in uninterruptible sleep on IO or hung NFS mounts, and consider terminating the systemd user with loginctl terminate-user or investigating tmux and screen sessions before killing to avoid data loss. Tools worth knowing Use `who` and `w` to find logged users, `ps` and `pstree` to inspect processes, `pkill` and `kill` to send signals, `loginctl` on systemd systems to control user realms, and `wall`, `write`, `lsof` or `fuser` to discover open files and warn users. Finish and next steps Safe termination is a small skill with big impact; practise notification, graceful signals, and measured escalation, and then deepen system knowledge by studying for Linux certifications like CompTIA Linux+ or LPIC-1 with intensive exam preparation at bitsandbytes.academy. Join Bits & Bytes Academy First class LINUX exam preparation. processes utilities security troubleshooting scripting
