Linux Systems

Understand what a Linux process actually is, how fork and exec create one, the five process states, and why a process stuck in D state cannot be killed.

Learn the signal table every SRE needs, why SIGKILL loses data, how graceful termination actually works, and how to keep work alive after you disconnect.

Read free -h correctly, tell RSS from VSZ, understand why Linux reports almost no free memory on a healthy host, and predict which process the OOM killer takes first.

Read and set the permission bits with confidence, understand what setuid and the sticky bit actually do, and know where identity lives on a Linux host.

Work both major package families with confidence, find out which package owns a file, and understand why an unpinned upgrade is a production change.

Cross the user/kernel boundary deliberately: read file descriptor tables, navigate /proc, and use strace to find out what a hung process is actually waiting on.

Write a correct unit file, understand why a crash-looping service eventually stops restarting itself, and query the journal precisely enough to find the first failure.

Follow a host from power-on to a running service, know which log holds which stage, and diagnose the boot that hangs before anything you recognise appears.

Read the kernel ring buffer like a log, tune the kernel with sysctl in a way that survives reboot, and recognise the hardware and driver failures that never reach an application log.

Resolve the classic df-versus-du disagreement, diagnose inode exhaustion, and read I/O metrics well enough to tell a slow disk from a busy one.

Turn a million-line log into an answer with grep, awk and sort, understand where logs come from and where they go, and stop a runaway log from filling the disk.

Walk the stack from link to application in a fixed order, read socket states correctly, and trace the DNS resolution path that quietly causes most intermittent latency.

Lay out storage that can grow, choose between RAID levels on their real trade-offs, and tune mount options and schedulers for the workload you actually have.

Close the doors that matter on a production host — SSH configuration, packet filtering, mandatory access control, and dropping root privilege without breaking the service.

Assemble everything into a repeatable first minute on an unhealthy host: what load average really counts, how to tell saturation from utilisation, and how to name the bottleneck with evidence.