Using STRACE to troubleshoot issues
summary
strace shows every syscall and timestamps, used by Red Hat support to diagnose userland failures; recommended flags are -f -tt -v -y -s 1024 to follow forks, decode strings and get ms timestamps. six real cases include slow df (400 ms per failed file lookup), slow SSH where the server waits 25s on a systemd session call, tracing services by attaching to PID 1, and boot-wide tracing; note strace fails for pure CPU spins or in-kernel hangs.
AI-written summary · about
chapters (5)
- 0:00Start
- 4:22How does strace work?
- 8:55Recommended usage
- 17:26stracing a command slowness
- 30:20stracing boot activity
