305-300 LPIC-3: Virtualization and Containerization - Exam 305, version 3.0 Questions and Answers

 A Pod in Kubernetes is a collection of one or more containers that share the same network and storage resources, and a specification for how to run the containers. A Pod is the smallest unit of workload Kubernetes can run, meaning that it cannot be divided into smaller units. Therefore, option C is correct. All containers of a Pod run on the same node, which is the smallest unit of computing hardware in Kubernetes. A node is a physical or virtual machine that hosts one or more Pods. Therefore, option A is also correct. Pods are not always created automatically and cannot be explicitly configured. Pods can be created manually using YAML or JSON files, or using commands like kubectl run or kubectl create. Pods can also be created automatically by higher-level controllers, such as Deployment, ReplicaSet, or StatefulSet. Therefore, option B is incorrect. When a Pod fails, Kubernetes does not restart the Pod on another node by default. Pods are ephemeral by nature, meaning that they can be terminated or deleted at any time. If a Pod is managed by a controller, the controller will create a new Pod to replace the failed one, but it may not be on the same node. Therefore, option D is incorrect. systemd is not used to manage individual Pods on the Kubernetes nodes. systemd is a system and service manager for Linux operating systems that can start and stop services, such as docker or kubelet. However, systemd does not interact with Podsdirectly. Pods are managed by the kubelet service, which is an agent that runs on each node and communicates with the Kubernetes control plane. Therefore, option E is incorrect. References:

Pods | Kubernetes

What is a Kubernetes pod? - Red Hat

What’s the difference between a pod, a cluster, and a container?

What are Kubernetes Pods? | VMware Glossary

Kubernetes Node Vs. Pod Vs.Cluster: Key Differences - CloudZero

Capabilities are a way of implementing fine-grained access control in Linux. They are a set of flags that define the privileges that a process can have. By default, a process inherits the capabilities of its parent, but some capabilities can be dropped or added by the process itself or by the kernel. In the context of container virtualization, capabilities are used to prevent processes from performing actions that might infringe the container, such as accessing the host’s devices, mounting filesystems, changing the system time, or killing other processes. Capabilities allow containers to run with a reduced set of privileges, enhancing the security and isolation of the container environment. For example, Docker uses a default set of capabilities that are granted to the processes running inside a container, and allows users to add or drop capabilities as needed12. References:

Capabilities | Docker Documentation1

Linux Capabilities: Making Them Work in Containers2

 Libvirt is a toolkit that provides a common API for managing different virtualization technologies, such as KVM, Xen, LXC, and others. Libvirt allows users to configure and control various aspects of a virtual machine (also called a domain), such as its CPU, memory, disk, network, and other resources. Among the resources that can be limited by libvirt for a KVM domain are:

Amount of CPU time: Libvirt allows users to specify the number of virtual CPUs (vCPUs) that a domain can use, as well as the CPU mode, model, topology, and tuning parameters. Users can also set the CPU shares, quota, and period to control the relative or absolute amount of CPU time that a domain can consume. Additionally, users can pin vCPUs to physical CPUs or NUMA nodes to improve performance and isolation. These settings can be configured in the domain XML file under the  and  elements12.

Size of available memory: Libvirt allows users to specify the amount of memory that a domain can use, as well as the memory backing, tuning, and NUMA node parameters. Users can also set the memory hard and soft limits, swap hard limit, and minimum guarantee to control the memory allocation and reclaim policies for a domain. These settings can be configured in the domain XML file under the , , and  elements13.

The other resources listed in the question are not directly limited by libvirt for a KVM domain. File systems allowed in the domain are determined by the disk and filesystem devices that are attached to the domain, which can be configured in the domain XML file under the  and  elements14. Number of running processes and number of available files are determined by the operating system and the file system of the domain, which are not controlled by libvirt.

References:

libvirt: Domain XML format

CPU Allocation

Memory Allocation

Hard drives, floppy disks, CDROMs

 Packer is a tool that automates the creation of identical machine images for multiple platforms from a single source configuration. Packer works by creating an instance based on a source image, which is a pre-existing image that serves as a starting point. Packer then connects to the instance through a network connection, such as SSH or WinRM, and runs various commands and scripts to install and configure software within the instance. Packer then shuts down the instance and creates a new system image from it, which can be used to launch new instances. Packer supports many platforms, such as AWS, Azure, VMware, Docker, and others. Packer does not install any software or run any daemon within the target image, nor does it periodically connect to the running instances to re-apply the template. Packer also does not modify the source image directly, but creates a new image from the modified instance. References:

Packer by HashiCorp

HashiCorp Packer - Build Automated Machine Images

Introduction | Packer | HashiCorp Developer

The virt-diff command-line tool can be used to list the differences between files in two virtual machines or disk images. The output shows the changes to a virtual machine’s disk images after it has been running. The command can also be used to show the difference between overlays1. To specify two guests, you have to use the -a or -d option for the first guest, and the -A or -D option for the second guest. For example: virt-diff -a old.img -A new.img1. Therefore, the correct command to list all differences between the disk images vm1-snap.img and vm1.img is: virt-diff -a vm1-snap.img -A vm1.img. The other commands are not related to finding differences between disk images. virt-delta is a tool to create delta disks from two disk images2. virt-cp-in is a tool to copy files and directories into a virtual machine disk image3. virt-cmp is a tool to compare two files or directories in a virtual machine disk image4. virt-history is a tool to show the history of a virtual machine disk image5. References:

21.13. virt-diff: Listing the Differences between Virtual Machine Files …

21.14. virt-delta: Creating Delta Disks from Two Disk Images …

21.6. virt-cp-in: Copying Files and Directories into a Virtual Machine Disk Image …

21.7. virt-cmp: Comparing Two Files or Directories in a Virtual Machine Disk Image …

21.8. virt-history: Showing the History of a Virtual Machine Disk Image …

 LXC and Docker are both container technologies that use Linux kernel features to create isolated environments for running applications. The main mechanisms that they use are:

Linux Capabilities: These are a set of privileges that can be assigned to processes to limit their access to certain system resources or operations. For example, a process with the CAP_NET_ADMIN capability can perform network administration tasks, such as creating or deleting network interfaces. Linux capabilities allow containers to run with reduced privileges, enhancing their security and isolation.

Kernel Namespaces: These are a way of creating separate views of the system resources for different processes. For example, a process in a mount namespace can have a different file system layout than the host or other namespaces. Kernel namespaces allow containers to have their own network interfaces, process IDs, user IDs, and other resources, without interfering with the host or other containers.

Control Groups: These are a way of grouping processes and applying resource limits and accounting to them. For example, a control group can limit the amount of CPU, memory, disk I/O, or network bandwidth that a process or a group of processes can use. Control groups allow containers to have a fair share of the system resources and prevent them from exhausting the host resources.

POSIX ACLs and file system permissions are not mechanisms used by LXC and Docker to create containers. They are methods of controlling the access to files and directories on a file system, which can be applied to any process, not just containers.

References:

LXC vs Docker: Which Container Platform Is Right for You?

LXC vs Docker: Why Docker is Better in 2023 | UpGuard

What is the Difference Between LXC, LXD and Docker Containers

lxc - Which container implementation docker is using - Unix & Linux Stack Exchange

The command lxc-checkpoint is used to checkpoint and restore containers. Checkpointing a container means saving the state of the container, including its memory, processes, file descriptors, and network connections, to a file or a directory. Restoring a container means resuming the container from the saved state, as if it was never stopped. Checkpointing and restoring containers can be useful for various purposes, such as live migration, backup, debugging, or snapshotting. The command lxc-checkpoint has the following syntax:

lxc-checkpoint {-n name} {-D path} [-r] [-s] [-v] [-d] [-F]

The options are:

-n name: Specify the name of the container to checkpoint or restore.

-D path: Specify the path to the file or directory where the checkpoint data is dumped or restored.

-r, --restore: Restore the checkpoint for the container, instead of dumping it. This option is incompatible with -s.

-s, --stop: Optionally stop the container after dumping. This option is incompatible with -r.

-v, --verbose: Enable verbose criu logging. Only available when providing -r.

-d, --daemon: Restore the container in the background (this is the default). Only available when providing -r.

-F, --foreground: Restore the container in the foreground. Only available when providing -r.

The command lxc-checkpoint uses the CRIU (Checkpoint/Restore In Userspace) tool to perform the checkpoint and restore operations. CRIU is a software that can freeze a running application (or part of it) and checkpoint it to a hard drive as a collection of files. It can then use the files to restore and run the application from the point it was frozen at1.

The other statements about the command lxc-checkpoint are not correct. It does not create a clone or an image of a container, nor does it double the memory consumptionof the container. It can work on both running and stopped containers, depending on the options provided. References:

Linux Containers - LXC - Manpages - lxc-checkpoint.12

lxc-checkpoint(1) - Linux manual page - man7.org3

CRIU4

Resource management for full virtualization is the process of allocating and controlling the physical resources of the host system to the virtual machines running on it. The hypervisor is the software layer that performs this task, by providing each virtual machine with a virtual hardware of a defined capacity that limits the resources of the virtual machine. For example, the hypervisor can specify how many virtual CPUs, how much memory, and how much disk space each virtual machine can use. The hypervisor can also enforce resource isolation and prioritization among the virtual machines, to ensure that they do not interfere with each other or consume more resources than they are allowed to. The hypervisor cannot provide fine-grained limits to internal elements of the guest operating system, such as the number of processes, because the hypervisor does not have access to the internal state of the guest operating system. The guest operating system is responsible for managing its own resources within the virtual hardware provided by the hypervisor. For example, the guest operating system can create an arbitrary amount of network sockets, as long as it does not exceed the network bandwidth allocated by the hypervisor. Full virtualization can pose limits to virtual machines, and does not always assign the host system’s resources in a first-come-first-serve manner. The hypervisor can use various resource management techniques, such as reservation, limit, share, weight, and quota, to allocate and control the resources of the virtual machines. The hypervisor can also use resource scheduling algorithms, such as round-robin, fair-share, or priority-based, to distribute the resources among the virtual machines according to their needs and preferences. All processes created within the virtual machines are not transparently and equally scheduled in the host system for CPU and I/O usage. The hypervisor can use different scheduling policies, such as proportional-share, co-scheduling, or gang scheduling, to schedule the virtual CPUs of the virtual machines on the physical CPUs of the host system. The hypervisor can alsouse different I/O scheduling algorithms, such as deadline, anticipatory, or completely fair queuing, to schedule the I/O requests of the virtual machines on the physical I/O devices of the host system. The hypervisor can also use different resource accounting and monitoring mechanisms, such as cgroups, perf, or sar, to measure and report the resource consumption and performance of the virtual machines. References:

Oracle VM VirtualBox: Features Overview

Resource Management as an Enabling Technology for Virtualization - Oracle

Introduction to virtualization and resource management in IaaS | Cloud Native Computing Foundation

IaaS is a type of cloud computing service that offers essential compute, storage, and networking resources on demand, on a pay-as-you-go basis. IaaS is one of the four types of cloud services, along with software as a service (SaaS), platform as a service (PaaS), and serverless12. IaaS eliminates the need for enterprises to procure, configure, or manage infrastructure themselves, and they only pay for what they use23. Some examples of IaaS providers are Microsoft Azure, Google Cloud, and Amazon Web Services.

The command that deletes all volumes which are not associated with a container is docker volume prune. This command removes all unused local volumes, which are those that are not referenced by any containers. By default, it only removes anonymous volumes, which are those that are not given a specific name when they are created. To remove both unused anonymous and named volumes, the --all or -a flag can be added to the command. The command will prompt for confirmation before deleting the volumes, unless the --force or -f flag is used to bypass the prompt. The command will also show the total reclaimed space after deleting the volumes12.

The other commands listed in the question are not valid or do not have the same functionality as docker volume prune. They are either made up, misspelled, or have a different purpose. These commands are:

docker volume cleanup: This command does not exist in Docker. There is no cleanup subcommand for docker volume.

docker volume orphan -d: This command does not exist in Docker. There is no orphan subcommand for docker volume, and the -d flag is not a valid option for any docker volume command.

docker volume vacuum: This command does not exist in Docker. There is no vacuum subcommand for docker volume.

docker volume garbage-collect: This command does not exist in Docker. There is no garbage-collect subcommand for docker volume.

References:

docker volume prune | Docker Docs

How to Remove all Docker Volumes - YallaLabs.

QEMU can provide some network services in a user network, which is a mode of networking that does not require any administrator privilege to run. The user network uses the SLIRP TCP/IP emulator to create a virtual NAT’ted subnet, with a DHCP server started by QEMU that gives out IP addresses to the guest machines and puts the host on 10.0.2.21. QEMU can also provide a TFTP server in the user network, which can be used to boot the guest machines from a network image. The TFTP server can be configured with the -tftp option2. QEMU does not provide BGP, CIFS, or AppleTalk services in the user network. BGP is a routing protocol that is used to exchange routing information between autonomous systems on the Internet3. CIFS is a file-sharing protocol that is used to access files and printers on a network4. AppleTalk is a deprecated network protocol suite that was used by Apple devices5. These services require more advanced networking features than the user network can offer, such as bridging, routing, or tunneling.

References:

Documentation/Networking - QEMU

QEMU/Networking - Wikibooks, open books for an open world

Border Gateway Protocol - Wikipedia

Common Internet File System - Wikipedia

AppleTalk - Wikipedia

The output of xl list shows the state of the domains. The domain with ID 6, Ubuntu, has a state of “b—”. This means that the domain is blocked, which means it is idle or waiting for I/O.

https://xenbits.xen.org/docs/unstable/man/xl.1.html

Containerd is an industry-standard container runtime that uses Runc (a low-level container runtime) by default, but can be configured to use others as well1. Containerd manages the complete container lifecycle of its host system, from image transfer and storage to containerexecution and supervision1. It supports the standards established by the Open Container Initiative (OCI)1. Containerd does not require the Docker engine and Docker CLI to be installed, as it can be used independently or with other container platforms2. Containerd is not a text file format, nor does it run in each Docker container or provide DHCP client functionality. Containerd is not the initial process run at the start of any Docker container, as that is the role of the container runtime, such as Runc3. References: 1 (search for “containerd”), 2 (search for “Containerd is an open source”), 3 (search for “It uses rune to start containers”).

Cloud-init is a tool that processes configurations and runs through five stages during the initial boot of Linux VMs in a cloud. It allows users to customize a Linux VM as it boots for the first time, by applying user data to the instance. User data can include scripts, commands, packages, files, users, groups, SSH keys, and more. Cloud-init can also interact with various cloud platforms and services, such as Azure, AWS, OpenStack, and others. The purpose of cloud-init is to prepare the generic image of an laaS instance to fit a specific instance’s configuration, such as hostname, network, security, and application settings. References:

Cloud-init - The standard for customising cloud instances

Understanding cloud-init - Azure Virtual Machines

Tutorial - Customize a Linux VM with cloud-init in Azure - Azure Virtual Machines

The file procs in a cgroup directory contains the list of processes belonging to this cgroup. Each line in the file shows the PID of a process that is a member of the cgroup. A process can be moved to a cgroup by writing its PID into the cgroup’s procs file. For example, to move the process with PID 24982 to the cgroup cg1, the following command can be used: echo 24982 > /sys/fs/cgroup/cg1/procs1. The file procs is different from the file tasks, which lists the threads belonging to the cgroup. The file procs can be used to move all threads in a thread group at once, while the file tasks can be used to move individual threads2. References:

Creating and organizing cgroups · cgroup2 - GitHub Pages

Control Groups — The Linux Kernel documentation