Skip to content

Course

Linux Systems Administration

Run Linux servers like an operator: systemd units that restart right, storage that survives reboots, firewalls that actually filter, and a calm first hour of an incident.

9 lessons

Course Q&A

Running services with systemd

Units, journalctl, and the sandboxing directives that keep a compromised daemon boxed in.

  1. 01Reading a systemd unit file like an operatorWhere units live, what the three sections mean, and why systemctl edit beats editing files you don't own.18mOpen
  2. 02journalctl: finding the one line that mattersThe flags that cover real queries, making the journal persistent, and reading around a crash or OOM kill.20mOpen
  3. 03Sandboxing a service with systemd directivesDynamicUser, ProtectSystem, CapabilityBoundingSet — shrink a daemon's blast radius with a few lines of INI.25mMembers

Storage and networking that survive reboots

fstab that boots, LVM that grows online, and firewall rules that actually filter traffic.

  1. 01fstab, mounts, and the empty-mountpoint disasterUUID-based fstab that survives reboots, findmnt verification, and stopping writes to unmounted directories.20mMembers
  2. 02Growing storage without downtime: LVM, ext4, and XFSPV-VG-LV-filesystem growth order, lvextend -r, cloud disk expansion, and why shrinking is the dangerous direction.25mMembers
  3. 03What's actually listening: ip, ss, and a real firewallMap sockets to processes with ss, dodge bind-address traps, write policy-drop nftables, and spot the Docker bypass.25mMembers

Hardening, logging, and incident basics

sshd lockdown, logs that outlive the attacker, and a field-tested triage order for hour one.

  1. 01Locking down sshd and local accountsKey-only auth, AllowGroups, sudoers hygiene, and the reload protocol that keeps you from locking yourself out.25mPremium
  2. 02Logs that outlive the attacker: remote syslog and auditdShip logs off-box with rsyslog, rotate sanely with logrotate, and watch high-value files with auditd.25mPremium
  3. 03The first hour of a Linux incidentCapture volatile evidence first, hunt persistence, isolate without powering off, and never trust the box's tools.30mPremium