Candidates who can start a container still lose marks on what holds it in place, and on which tool owns a connection. Set 2 aims at exactly those two gaps.
LPIC-3 305 — Three virtualization terms the exam expects you to keep separate. Watch on YouTube · CertCrafter What do these free LPIC-3 305 questions cover?
Three questions, three habits. One asks you to read the state of a container host, one asks what a storage line says about how images sit on disk, and one asks what libvirt does when you hand it a connection URI. None is a definition question. Each gives you a command or a line of output and asks what follows from it.
The first set is on a separate page, so you can do both in a sitting. Take every question cold: write your answer down, then open the explanation. The channel also has a short clip on three virtualization terms the exam expects you to keep separate if you want a warm-up.
How does Linux actually confine a container?
A container is not a small virtual machine. It is a normal process on the host kernel, fenced in by several separate mechanisms. LPI’s objective for container concepts names them: kernel namespaces, control groups, capabilities, and the role of seccomp, SELinux and AppArmor. It also says LXC and Docker both use these.
Each layer answers a different question, and that is the whole trick. When a stem describes a symptom, ask which layer could cause it.
| Mechanism | What it limits | A command or path to know |
|---|---|---|
| Namespaces | What the process can see | unshare, nsenter, /proc/PID/ns |
| Control groups | How much it can use | /sys/fs/cgroups |
| Capabilities | Which root powers it keeps | capsh |
| seccomp | Which system calls it may make | named in 352.1 |
| SELinux, AppArmor | Which objects a policy lets it touch | profiles loaded per host |
The listed utilities come straight from the objective. Anyone who has run unshare by hand and looked at the entries under /proc has an easier time than anyone who has only read about it.
What do administrators check first on a confined host?
In practice you rarely start from theory. A container cannot write somewhere, or a service that works on one host fails on another, and the question is whether a profile is involved. The first move is to see what is loaded and what is enforced, not to disable the profile and hope.
Teams that skip that step end up turning confinement off to get a deploy through, then leaving it off. A better habit is a short checklist per incident: which profile, which mode, which process. The first free question below is built around that habit, and it expects you to know which tool reports state and which tools merely change it.
Which Docker facts does the exam expect you to read from output?
Docker carries 9 of the 60 weighted points, tied with libvirt for the heaviest objective. LPI’s list for it includes dockerd, /etc/docker/daemon.json and /var/lib/docker/, plus images, volumes, logging, networking and Dockerfiles. Notice how many of those are places on a filesystem. The exam likes to hand you a line from a command and ask what it implies.
On a real build runner the same skill saves an afternoon. When a disk fills or a pull is slow, reading the storage driver and the backing filesystem tells you where to look before you touch anything. The second free question below is about that reading.
How does libvirt choose which hypervisor to talk to?
Objective 351.4 is the other 9-point objective. LPI asks you to understand how libvirt interacts with Xen and QEMU, to manage connections and nodes, and to create and manage domains, storage pools, volumes and virtual networks. The files and tools it lists are libvirtd, /etc/libvirt/ and virsh with its subcommands.
One idea ties much of this together: the connection decides which driver does the work. The same virsh binary can manage different hypervisors, and on a mixed estate that is exactly why a command that looked right ran against the wrong host. Before you run anything destructive, check which connection you are on. The third free question tests whether you know the order of events.
Where do candidates lose marks on these topics?
Two patterns show up again and again. The first is mixing up tools that report state with tools that change it, which costs marks wherever a stem asks what a command does and offers five names that look alike. The second is reasoning by analogy: treating a container like a virtual machine, or assuming that a driver is chosen by what is installed rather than by the connection.
Both fixes are practical. Run the commands on a lab host, read the output, and break something on purpose. LPI says the objectives should sit at the core of any study plan, so keep them open beside the terminal and tick off what you have run. The LPIC-3 305 common mistakes guide covers more habits like these, and the command walkthroughs show a command from start to finish.
For the full fact sheet see the LPIC-3 305-300 guide, and the hub of 305-300 guides lists the rest.
What these three questions will not do
They are not copied from a live exam. They are written from LPI’s published objectives, and three questions cannot tell you whether you are ready for 60 in 90 minutes. They show whether a habit is sound.
The LPIC-3 305-300 practice test bundle picks up where these stop, with six tests and 720 questions, every answer explained. You can also try the free scored quiz. Next step: open the first question and commit to an answer before you read the explanation.
Comments
Comments are read by a person before they appear. How we moderate.