Free XK0-006 practice test: 30 questions with answers

find combines a size test (+100M means larger than 100 MB) with a modification-time test (-mtime -1 means within the last day). ls and du do not filter by both criteria, and locate uses a database that does not check modification time this way.

du with -x stays on one filesystem, and sort -h orders the human-readable sizes so the largest directories appear last. fstab lists configured mounts, mount -a mounts them, and lsblk shows block devices rather than per-directory usage.

Restart=on-failure tells systemd to restart the service when it exits abnormally. WantedBy controls boot-time enablement, After controls ordering, and User sets the account that runs the process, so none of those restart a crashed service.

Group membership is read at login, so a session started before the user was added to devs will not have write permission through that group. The setgid bit in mode 2775 is correct for a shared directory, and the sticky bit is not what grants write access.

lvextend grows the logical volume, and the -r option or a following resize2fs grows the ext4 filesystem online. Deleting or reformatting the volume destroys data, and changing the physical partition alone does not extend the logical volume.

dnf provides searches the package metadata for the package that owns a file path, which answers the question without installing. rpm -e removes a package, systemctl manages services, and apt purge removes packages on Debian-based systems.

timedatectl reports whether NTP synchronization is active, and chronyc tracking shows the selected source and offset when chrony is in use. lsof, nmcli and cron checks do not describe time synchronization.

SIGTERM asks the process to terminate, which allows it to clean up and save state. SIGKILL cannot be caught and skips cleanup, SIGSTOP pauses rather than ends the process, and SIGHUP usually means reload or hangup, not termination.

The sticky bit on a directory restricts deletion so that only the file owner, the directory owner or root can remove files, which is the typical protection for shared directories such as /tmp. The setuid and setgid bits affect execution and group inheritance, and execute permission does not control deletion.

A narrow sudoers rule, edited with visudo to check syntax, grants only the named command with elevated rights. Sharing the root password or granting a root shell gives far more access than needed.

sshd_config can restrict users or source addresses, and a host firewall rule limits connections to the management subnet, which enforces the requirement. Changing the port is obscurity, Telnet is insecure, and a public IP does not restrict who connects.

Mode 600 gives read and write only to the owner and no access to group or others. 644 lets others read, 660 lets the group write, and 777 gives everyone full access.

Service accounts should not have interactive shells, and nologin prevents login while keeping the account for its services. A longer password or added privileges increase risk, and enabling root login over SSH is unsafe.

ss -tnp shows established TCP connections with the owning process, and /proc/PID/exe links to the executable path. df, uptime and the message of the day do not show network process details.

SELinux denials are recorded as AVC messages in the audit log, and ausearch -m avc or sealert summarize them with the context and suggested fix. The other commands show host names, inode usage and scheduled jobs.

A default-deny zone with explicit allowances for the https and ssh services enforces the stated policy. Allowing everything and blocking one port does not meet the requirement, and relying on a router alone leaves the host open.

set -e exits when a command fails and set -u treats unset variables as errors, which makes scripts fail fast and predictably. set -x only traces commands, exporting PATH changes the search path, and chmod affects permissions rather than error handling.

podman ps and docker ps -a list containers, the second including stopped ones. Creating a volume, enabling the Docker service and archiving the container storage directory do not list containers.

A while read loop redirected from the file processes each line in turn, and -r keeps backslashes literal. The for loop in the distractor iterates over the file name, and the other options print or pipe without iterating over the lines.

A multi-stage build compiles in one stage and copies only the artifact into a small runtime image, which keeps build tools and intermediate files out. Installing build tools in the final image, large base images and baked-in secrets all increase size or risk.

A systemd unit or Quadlet definition lets the container start at boot and restart on failure under systemd supervision. A message-of-the-day entry, an open terminal session and a single reboot cron entry do not supervise the service.

Checking exit codes and exiting non-zero on failure makes the script's result reliable for automation and CI. Hiding errors, backgrounding without waiting and printing success early all hide real failures.

Cron runs jobs with a minimal environment, and a limited PATH often hides commands that exist in an interactive shell. Using full paths or setting PATH in the crontab fixes the issue. Python is supported by cron, and cron does not need a graphical session or special directories.

iostat reports device utilization and wait times, and iotop or pidstat -d show per-process I/O, which together locate the source. Address, connection and hostname commands do not measure disk activity.

getent hosts and dig test whether the name resolves through the configured resolvers, and checking resolver settings shows which servers are used. Cables, kernel reinstalls and shell changes do not affect DNS resolution.

systemctl status and journalctl show the error messages the service logged, and many daemons provide a syntax test mode that reveals configuration mistakes. Restarting blindly, deleting logs and disabling security controls do not identify the cause.

df -i reports inode totals and usage per filesystem, so a full inode table is visible even when block space remains. df -h shows block space only, free reports memory, and uptime reports load and time.

ip route show lists the routing table, and a missing default route explains why traffic leaving the subnet fails. The hostname, block devices and cron entries are unrelated to routing.

The kernel log records OOM killer decisions, including the process and memory usage, and journalctl -k shows those kernel messages with timestamps. Shell history, hosts entries and the message of the day do not contain this information.

A read-only mount prevents deletion regardless of ownership, and the immutable attribute set with chattr blocks changes even for owners. File name length, prompts and future timestamps do not block deletion.