LPIC-2 Block Devices, Filesystems and Advanced Storage Practice Test 3 — 15 Questions

Question 1

A storage administrator is preparing a new 10 TB LUN for a database server. The workload performs many small random writes, and the administrator wants to minimize write amplification and metadata overhead while keeping crash consistency. Which mkfs.xfs invocation is most appropriate?

Choose an answer to begin — your selection is scored in the full session.

Accepted Answer

mkfs.xfs -f -b size=4096 -m crc=1 -d agcount=4 /dev/sdc. For a database doing small random writes, a 4 KB block size matches the typical I/O size, CRC metadata checksums improve crash consistency, and a small allocation group count reduces metadata overhead on a single large LUN. The other invocations either use unsupported block sizes, disable checksums, or create excessive allocation groups.

Answer

mkfs.xfs -f -b size=512 -m crc=0 -d agcount=64 /dev/sdc

Answer

mkfs.xfs -f -b size=65536 -m crc=1 -d agcount=8 /dev/sdc

Answer

mkfs.xfs -f -b size=4096 -m crc=0 -d agcount=64 /dev/sdc

Question 2

A storage administrator is preparing a new 10 TB LUN for a database server. The workload performs many small random writes, and the administrator wants to minimize write amplification and metadata overhead while keeping crash consistency. Which mkfs.xfs invocation is most appropriate?

Choose an answer to begin — your selection is scored in the full session.

Accepted Answer

Ensure the multipath device maps are listed in the filter in /etc/lvm/lvm.conf with an accept rule and the underlying paths rejected.. When a LUN is visible over multiple paths, LVM by default may scan each underlying SCSI device and create duplicate PVs. The correct fix is to configure the devices/filter in lvm.conf so that only the multipath device (/dev/mapper/mpath*) is accepted and the individual paths are rejected. This keeps multipath management intact while giving LVM a single device to use.

Answer

Add the LUN's WWID to the blacklist in /etc/multipath.conf so multipathd ignores it.

Answer

Set multipath to 'failover' mode and disable the second path in /etc/multipath.conf.

Answer

Run pvcreate directly on both /dev/sd* paths and then use vgmerge to combine them.

Question 3

A technician must create a filesystem on a 4 TB LUN presented as /dev/mapper/mpatha. The filesystem will hold large video files, so extents of 64 MB are desired, and the mount point must survive a power loss without requiring a full e2fsck. Which command should the technician run?

Accepted Answer

mkfs.xfs -b size=4096 -d su=64m,sw=1 -L video /dev/mapper/mpatha. XFS is the appropriate choice for very large filesystems holding large files, because it uses B-tree indexed extents and recovers through journal replay instead of a full check after an unclean shutdown. Setting the block size to the page size and aligning the data stripe unit with the underlying storage geometry produces efficient allocation for large sequential workloads.

Answer

mkfs.xfs -b size=65536 -d su=64m,sw=1 /dev/mapper/mpatha

Answer

mkfs.btrfs -n 65536 -d raid0 -m raid1 /dev/mapper/mpatha

Answer

mkfs.ext4 -b 4096 -E stride=16,stripe-width=16 /dev/mapper/mpatha

Question 4

A file server exports a 4 TB directory over NFS to a render farm. Users report that large frame files written to the export occasionally appear truncated after a power event on the server. The export is configured in /etc/exports as /render farm.example.com(rw,async,no_subtree_check). Which change best protects data integrity across such crashes?

Accepted Answer

Replace async with sync in the export options and reload the export with exportfs -ra. The reported truncation after a power event points to the NFS async export option, which allows the server to acknowledge writes before they are committed to stable storage. Changing the export to sync makes the server wait for data to reach disk before replying, protecting integrity. The change is applied live with exportfs -ra, which re-reads /etc/exports and updates the kernel export table without remounting clients.

Answer

Add the sync option to the underlying ext4 mount in /etc/fstab and remount it with mount -o remount,sync

Answer

Add the no_wdelay option and remount the export with exportfs -v

Answer

Replace no_subtree_check with subtree_check to force directory validation on every write

Question 5

An administrator is migrating a server from ext4 to XFS. Before the migration, the administrator must verify which XFS features are available in the running kernel and ensure the filesystem can be mounted with project quotas. Which two commands or checks are appropriate? (Choose two.)

Accepted Answer

grep xfs /proc/filesystems. Verifying XFS availability before a migration requires checking the running kernel's filesystem support and inspecting the existing XFS filesystem's feature flags. Reading /proc/filesystems confirms the kernel driver is present, while xfs_info on the mount point reveals enabled features such as project quotas. Destructive or ext-only tools do not serve this verification purpose.

Answer

xfs_repair -n /dev/sdd

Answer

tune2fs -l /dev/sdd

Answer

mkfs.xfs -m crc=1 -d agcount=8 /dev/sdd

Question 6

An administrator is tuning an ext4 filesystem on a busy mail server and wants to reduce fragmentation and improve write performance. Which two mount options should the administrator consider? (Choose two.)

Accepted Answer

data=writeback. Reducing metadata writes and relaxing ordering constraints are the two levers that most affect ext4 write performance and fragmentation on a busy server. noatime eliminates access-time updates, and data=writeback reduces ordering overhead for data writes. The other options either affect SSD trimming, journal integrity checking, or barrier safety rather than the stated performance goals.

Answer

discard

Answer

journal_checksum

Answer

nobarrier

Question 7

An administrator needs to mount an XFS filesystem on /dev/vg_app/lv_logs at /var/log/app with the noatime option and have it persist across reboots. The filesystem uses a UUID. Which /etc/fstab entry correctly accomplishes this?

Accepted Answer

UUID=abcd-1234 /var/log/app xfs defaults,noatime 0 0. A persistent XFS mount by UUID requires the UUID= form in the device field, xfs as the type, the desired mount options, and fsck pass 0 because XFS does not use the traditional fsck pass ordering. Using /dev/vg_app/lv_logs would work but does not meet the UUID requirement, and declaring ext4 or setting fsck pass 1 or 2 is incorrect for XFS.

Answer

UUID=abcd-1234 /var/log/app ext4 defaults,noatime 0 2

Answer

UUID=abcd-1234 /var/log/app xfs noatime 1 1

Answer

/dev/vg_app/lv_logs /var/log/app xfs noatime 0 2

Question 8

A storage engineer is documenting the behavior of LVM RAID (dm-raid) volumes on a production server. Which two statements about LVM RAID are correct? (Choose two.)

Accepted Answer

An LVM RAID1 logical volume can be created with lvcreate --type raid1 -m 1.. LVM RAID volumes are created with lvcreate --type raidN, and a two-way RAID1 uses -m 1. Recovery of a failed leg is performed with lvconvert --repair, which replaces the failed device and resynchronizes the array. These two statements reflect the documented creation and repair workflows for dm-raid volumes, distinguishing LVM RAID from mdadm-managed arrays.

Answer

LVM RAID volumes require the dm-raid kernel module and mdadm userspace tools.

Answer

LVM RAID automatically converts an existing linear LV to RAID5 when a third device is added.

Answer

LVM RAID volumes cannot be extended while active.

Question 9

An administrator wants /srv/data to be stored on a Btrfs subvolume named data inside filesystem /dev/sdb1, with automatic snapshots taken before each package upgrade. The root filesystem is ext4 and must remain so. Which combination of steps accomplishes this?

Accepted Answer

Create the Btrfs filesystem on /dev/sdb1, mount it at /mnt, run btrfs subvolume create /mnt/data, then add a mount entry referencing subvol=data for /srv/data.. A Btrfs subvolume is a separately mountable tree inside a Btrfs filesystem, created after the filesystem is mounted and referenced at mount time with the subvol= option. This allows the ext4 root to remain in place while the subvolume is exposed at the target directory. Snapshots are then taken directly against the subvolume path.

Answer

Run mkfs.btrfs --subvolume data /dev/sdb1 and mount /dev/sdb1 at /srv/data directly.

Answer

Format /dev/sdb1 with Btrfs, then run btrfs filesystem label data /dev/sdb1 and mount by label.

Answer

Create an LVM logical volume, format it with Btrfs, then use btrfs device add to attach /dev/sdb1 as a subvolume.

Question 10

An administrator prepares a new server with a hardware RAID controller whose cache lacks a battery backup unit, and the array will host a PostgreSQL database. Which TWO mkfs or mount choices are appropriate to protect data integrity on this array? (Choose two.)

Accepted Answer

Format with ext4 and mount with data=journal to serialize all data and metadata writes through the journal.. On storage with a volatile write cache, crash consistency depends on forcing cache flushes at the right moments, which write barriers provide. XFS enables barriers by default, and ext4 in data=journal mode commits data and metadata to the journal before final placement, so both preserve integrity across a power loss. Options that disable barriers trade that protection for speed.

Answer

Format with ext4 and mount with the discard option to let the controller manage flash wear leveling.

Answer

Format with ext4 and add the barrier=0 mount option to maximize write throughput.

Answer

Format with XFS and mount with the nobarrier option to reduce write latency.

Question 11

A system administrator notices that a new 1TB NVMe drive (/dev/nvme0n1) is not detected by the kernel. The hardware is confirmed working. Which troubleshooting step should be taken first to check if the drive is recognized by the system's PCI subsystem?

Accepted Answer

Run lspci to verify the NVMe controller is detected.. The NVMe drive is not detected by the kernel, but the hardware is confirmed working. The first step is to verify whether the PCI subsystem sees the NVMe controller, because NVMe devices are connected via the PCI Express bus. Running lspci lists all PCI devices, including the NVMe controller; if it does not appear, the issue is at the PCI or hardware level, not the block layer.

Answer

Run lsblk to list all block devices.

Answer

Check dmesg output for errors.

Answer

Run fdisk -l /dev/nvme0n1 to probe the drive.

Question 12

Which TWO statements about LVM thin provisioning are correct?

Accepted Answer

The filesystem on a thin volume must support the 'discard' option to free unused space.. Option B is correct because a thin volume's filesystem must be mounted with the 'discard' option (or periodically trimmed with fstrim) so that freed blocks are returned to the thin pool, otherwise the pool keeps reporting them as used. Option C is correct because thin provisioning's core feature is over-provisioning: the sum of thin volume virtual sizes can exceed the physical capacity of the thin pool, with space allocated on demand as data is written. Option A is wrong because thin pools are created with 'lvcreate --type thin-pool' (or 'lvconvert --thinpool'), while 'pvcreate' only initializes a physical volume for LVM use. Option D is wrong because dm-cache is a separate LVM caching target for improving performance, not a requirement for thin provisioning, which relies on the dm-thin target. Option E is wrong because thin volumes do not automatically grow; when the pool is exhausted, writes fail unless the pool is extended or the volume is resized manually.

Answer

Thin pools are created using the 'pvcreate' command.

Answer

Thin provisioning requires a dm-cache device to function.

Answer

Thin volumes automatically grow when they run out of space.

Question 13

The administrator wants to create a RAID 1 array using /dev/sdb1 and /dev/sdc1, then create a filesystem and mount it at /mnt/raid1. Which command sequence should be used first?

Accepted Answer

mdadm --create /dev/md0 --level=1 --raid-devices=2 /dev/sdb1 /dev/sdc1. The first step in creating a RAID 1 array is to assemble the underlying block devices into a RAID device using mdadm. The command `mdadm --create /dev/md0 --level=1 --raid-devices=2 /dev/sdb1 /dev/sdc1` initializes the RAID metadata and creates the array at /dev/md0. Only after the array is created can a filesystem be built on it and then mounted.

Answer

mkfs.ext4 /dev/md0

Answer

pvcreate /dev/sdb1 /dev/sdc1

Answer

fdisk /dev/sdb && fdisk /dev/sdc

Question 14

Order the steps to configure a Linux system as a DHCP client using dhclient.

Accepted Answer

Configure the interface for DHCP, then run dhclient, then check the IP address, then verify the default route, then test connectivity.. First configure the interface for DHCP, then run dhclient, check address, verify route, and test.

Answer

Run dhclient first, then configure the interface for DHCP, then check the IP address, then verify the default route, then test connectivity.

Answer

Check the IP address first, then configure the interface for DHCP, then run dhclient, then verify the default route, then test connectivity.

Answer

Configure the interface for DHCP, then check the IP address, then run dhclient, then verify the default route, then test connectivity.

Question 15

Arrange the steps to configure a Linux system as a PostgreSQL database server.

Accepted Answer

Install PostgreSQL, then initialize the database cluster, then start the PostgreSQL service, then create a user and database, then test the connection.. The correct order to configure a Linux system as a PostgreSQL database server is: install PostgreSQL, initialize the database cluster using initdb, start the PostgreSQL service, create a database user and database, and then test the connection. Each step depends on the previous one; for example, starting the service requires the cluster to exist, and creating users requires the service to be running.

Answer

Install PostgreSQL, then start the PostgreSQL service, then initialize the database cluster, then create a user and database, then test the connection.

Answer

Install PostgreSQL, then initialize the database cluster, then create a user and database, then start the PostgreSQL service, then test the connection.

Answer

Install PostgreSQL, then start the PostgreSQL service, then create a user and database, then initialize the database cluster, then test the connection.