Free 101-500 practice test: 30 questions with answers

!ssh is history expansion that reruns the latest command starting with ssh. !! reruns the previous command in full, $ssh expands a shell variable named ssh, and %ssh refers to a job in job control rather than the history.

head prints the first ten lines of its input by default. tail prints the last lines, tac prints lines in reverse order, and wc counts lines, words and bytes instead of showing the text.

uniq only collapses adjacent duplicate lines, so the input must be sorted first; uniq -c then prefixes each line with its count, and sort -rn orders the results from highest to lowest. Running uniq without sorting misses repeats that are not adjacent, grep -c . only counts non-empty lines, and wc -l gives a single total.

2> redirects file descriptor 2, standard error, to the file while standard output stays on the terminal. > sends only standard output to the file, &> sends both streams to the file, and < feeds the file's contents to standard input instead.

A pipe feeds standard output into the next command's standard input, and tee writes its input to a named file and to standard output at the same time. A pipe carries standard output by default rather than standard error, and its data passes through a kernel buffer rather than a temporary file on disk.

SIGKILL is enforced by the kernel and cannot be caught, blocked or ignored, so the process gets no chance to clean up. SIGTERM, SIGHUP and SIGINT can all be handled by a program, which is why they are often used to request a clean shutdown or a configuration reload.

pgrep searches the running process table by name and prints the PIDs of matching processes. uname -a shows kernel details, the which command only shows the path of an executable, and df -h reports disk space usage.

Niceness ranges from -20 (highest priority) to 19 (lowest), so nice -n 19 gives the backup the smallest CPU share. A negative value raises priority and requires root, time only reports how long a command took, and nohup only protects the job from hangup signals.

mkfs.ext4 writes a new ext4 filesystem onto the partition. fsck.ext4 checks an existing filesystem, fdisk edits partition tables rather than creating filesystems, and mount attaches a filesystem to the directory tree.

fsck writes repairs to on-disk structures, and if the kernel is writing to the same filesystem at the same time the metadata can be corrupted, which is why the filesystem should be unmounted or checked read-only. The other options describe behaviour that is wrong for fsck: it does not wipe user files, it is not limited to reading the superblock, and it does not change the runlevel.

/etc/fstab lists the devices, mount points and options that the system mounts at boot. /etc/mtab reflects what is currently mounted rather than configuring mounts, /etc/exports defines NFS shares for a server, and /etc/hosts maps host names to addresses.

750 gives the owner rwx (7), the group r-x (5) and others nothing (0). 640 removes execute for the owner and group so the directory cannot be entered, 755 gives other users read and enter access, and 700 removes all group access.

A hard link is another directory entry for the same inode, so the data is freed only after the last link is removed. The path-storing description fits a symbolic link instead, a hard link cannot span filesystems, and removing the original file does not remove its hard links.

/usr/local is the FHS location for software installed by the administrator outside package-managed trees. /usr/bin holds executables owned by the package manager, /bin holds essential system binaries, and /etc holds configuration rather than programs.

Partitions append their number to the whole-disk name, so the second partition of /dev/sda is /dev/sda2. /dev/sd2a and /dev/sda02 do not follow the naming scheme, and /dev/hda2 refers to a legacy IDE disk rather than a SATA disk.

df -i reports inode usage per filesystem, and a filesystem can refuse new files when its inodes are exhausted even with free blocks. df -h reports only block space, du -sh measures the size of a directory tree, and lsblk -f lists block devices with their filesystem details.

dpkg works only on the package file it is given and does not download missing dependencies, while apt consults the configured repositories to resolve and fetch them. apt can install a local .deb by path, and dpkg does not use the apt repository list at all.

dnf (or yum on older systems) installs a local .rpm path and resolves its dependencies from the enabled repositories. rpm -ivh installs the file without looking up missing dependencies, --nodeps skips the dependency check entirely, and rpm -e removes a package instead of installing one.

ldd prints each shared library a dynamic executable depends on and the path where the dynamic linker resolves it. ldconfig -v rebuilds the linker cache and does not inspect one executable, strip removes symbol information, and file only reports the file type.

The dynamic linker searches only the directories listed in /etc/ld.so.conf, including the files in /etc/ld.so.conf.d, and ldconfig rebuilds its cache, so adding the path there and running ldconfig makes the library visible. Copying to /usr/bin does not place it where the linker searches, chmod does not affect library loading, and PATH only controls where executables are found.

virtio provides paravirtualized block and network drivers that KVM guests use for efficient I/O. Emulated e1000 presents an emulated NIC with more overhead, ACPI handles power and platform events, and iSCSI is a storage protocol rather than a driver family for virtual hardware.

/var holds variable data such as logs and spool files, so a separate partition keeps its growth from filling /. /home holds user data, /opt holds add-on packages, and /srv holds data served by the system, and none of these is the usual home of growing logs.

apt-get update downloads fresh package indexes from the configured repositories and leaves installed software unchanged. apt-get upgrade installs newer versions of installed packages, dpkg --configure -a finishes configuring partly installed packages, and apt-get clean empties the local package cache.

UEFI firmware loads boot files from the EFI System Partition, a FAT-formatted partition that holds the bootloader binaries. The BIOS boot partition serves GRUB on GPT disks under legacy BIOS, swap is used for paging, and firmware does not read the root filesystem directly.

multi-user.target corresponds to runlevel 3, with networking and services running on a text console and no display manager. graphical.target maps to runlevel 5, rescue.target to single-user runlevel 1, and poweroff.target to halting the system, which is runlevel 0.

lspci lists devices attached to the PCI bus, where network and graphics adapters are normally connected. lsusb lists USB devices, lsmod lists loaded kernel modules, and lsblk lists block devices such as disks and partitions.

Firmware performs POST and hands control to a bootloader such as GRUB, which loads the kernel and initramfs, so B and D are correct. A is wrong because init runs only after the kernel has started, and C is wrong because boot menu entries are read by the bootloader while /etc/passwd holds user accounts.

The initramfs is a temporary root filesystem in memory that holds the drivers and tools needed to find and mount the real root filesystem. It does not hold user home directories, replace the kernel scheduler, or store the boot log, which are not its job.

set-default changes the target that systemd starts at boot, so the change persists across reboots. isolate multi-user.target switches to text mode immediately but leaves the boot default unchanged, set-default graphical.target would boot to the desktop, and start graphical.target launches the desktop for the current session only.

/proc/cpuinfo is a kernel-generated file that lists each processor's model name and feature flags. /proc/meminfo reports memory usage, /proc/partitions lists known block devices and their sizes, and /proc/version shows the kernel version string.