Processes and service lifecycles

A process has an identity, parent, environment, file descriptors, resource limits, and a lifecycle. Signals request an action, but a process may handle or ignore many signals.

Diagnose in layers

Use ps or pgrep to find the process, lsof to inspect open files and sockets, and top or vmstat to observe resource pressure. Check logs and service status before sending a terminating signal. Prefer a graceful stop, verify that the process exited, and escalate only when necessary.

Service managers

A service manager starts processes, tracks dependencies, restarts failures, and connects logs to the operating system. When a service fails, inspect the unit configuration, recent journal entries, environment, permissions, ports, and dependencies in that order. An automatic restart can hide a crash loop, so alert on repeated restarts rather than treating restart success as health.