Block Devices, Filesystems and Advanced Storage practice questions

both A and B

Why wrong: Incorrect. This option is not a command; it is a self-referential phrase. The correct answer is to select both A and C.

fdisk -l

Why wrong: Incorrect. `fdisk -l` lists partition tables but does not display UUIDs.

Run lsblk to list all block devices.

Why wrong: lsblk only lists devices already recognized by the kernel.

Check dmesg output for errors.

Why wrong: dmesg is useful after lspci to diagnose driver issues.

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

Why wrong: fdisk requires the device to be present.

Export the DRBD block device via NFS and use a load balancer.

Why wrong: NFS adds a single point of failure and complexity.

Primary/Secondary DRBD resource with manual failover using drbdadm.

Why wrong: Manual failover does not meet the automatic failover requirement.

DRBD with Heartbeat to automatically promote the secondary on primary failure.

Why wrong: This provides automatic failover but only one active node, not optimal for HA.

mount -o remount,rw /dev/sda1

Why wrong: May succeed only if filesystem is already clean.

fsck -y /dev/sda1

Why wrong: Automatically answers yes to repairs, potentially causing data loss.

e2fsck -p /dev/sda1

Why wrong: Attempts automatic repair, which can cause data loss.

Thin pools are created using the 'pvcreate' command.

Why wrong: Thin pools are created with lvcreate --type thin-pool.

Thin provisioning requires a dm-cache device to function.

Why wrong: dm-cache is a separate feature for caching.

Thin volumes automatically grow when they run out of space.

Why wrong: Auto-extension requires monitoring and manual intervention.

Convert lv_data to linear mapping on /dev/sda and /dev/sdb only.

Why wrong: This loses the benefit of striping and may not improve performance.

Increase the stripe size to 256KB to reduce I/O overhead.

Why wrong: This may not address the slow disk; the issue is at the device level.

Remove /dev/sdc from the volume group and reduce the stripe count to 2.

Why wrong: Reducing stripes reduces parallelism, may hurt performance further.

mkfs -t ext4 /dev/sdb1 && mount /dev/sdb1 /data

Why wrong: Mounts immediately but not persistently; no fstab entry.

blkid /dev/sdb1 && echo 'UUID=... /data ext4 defaults 0 2' >> /etc/fstab && mount -a

Why wrong: blkid only displays UUID, does not create filesystem.

mkfs.xfs /dev/sdb1 && echo '/dev/sdb1 /data xfs defaults 0 2' >> /etc/fstab && mount -a

Why wrong: Uses XFS instead of ext4.

Thin pools automatically extend when they reach 80% usage.

Why wrong: Auto-extension is not default; must be configured with dmeventd.

A thin pool cannot be extended once created.

Why wrong: Thin pools can be extended with lvextend.

It is only supported on SSDs.

Why wrong: Thin provisioning works on any block device.

mkfs.ext4 /dev/md0

Why wrong: Filesystem can only be created after RAID array exists.

pvcreate /dev/sdb1 /dev/sdc1

Why wrong: pvcreate is for LVM physical volumes, not software RAID.

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

Why wrong: Partitioning may be needed but does not create RAID array.

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

Why wrong: Running dhclient before configuring the interface for DHCP will fail because the interface is not set to use DHCP, so no lease will be obtained.

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

Why wrong: Checking the IP address before running dhclient is premature because no DHCP has been performed yet, so the address will be either missing or a static one.

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

Why wrong: Checking the IP address before running dhclient will show no DHCP-assigned address (or a stale one), as dhclient has not yet been executed to obtain a lease.

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

Why wrong: This is incorrect because the PostgreSQL service cannot start successfully unless the database cluster has been initialized first. Starting before init will fail.

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

Why wrong: This is incorrect because creating a user and database requires the PostgreSQL service to be running. The service must be started before you can create users or databases.

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

Why wrong: This is incorrect because you cannot start the service or create users before initializing the database cluster. Initialization must happen first, before any other step involving the database.

Logical Volume (LV): The smallest allocatable unit of space on a Physical Volume.

Why wrong: Incorrect — this describes a Physical Extent, not a Logical Volume.

Volume Group (VG): A virtual block device that can be used as a filesystem.

Why wrong: Incorrect — this describes a Logical Volume, not a Volume Group.

iptables: A tool for monitoring file system integrity.

Why wrong: Incorrect — file system integrity is typically monitored by AIDE or Tripwire.

nmap: A tool for banning IPs after failed login attempts.

Why wrong: Incorrect — this describes fail2ban, which blocks IPs based on log failures.

tcpdump: A tool for scanning open ports.

Why wrong: Incorrect — port scanning is a function of nmap, not tcpdump.

echo 'scan' > /proc/scsi/scsi

Why wrong: This command is deprecated and may not work on modern kernels.

blockdev --rereadpt /dev/sdc

Why wrong: This command re-reads the partition table, but does not rescan for new devices.

service scsi restart

Why wrong: There is no standard SCSI service; this command is invalid.

mkfs.ext4 --block-size 4096 -m 2 /dev/sdb1

Why wrong: The flag for block size is -b, not --block-size.

mkfs.ext4 -B 4096 -m 2 /dev/sdb1

Why wrong: -B is not a valid flag for mkfs.ext4; block size is set with -b.

mkfs.ext4 -b 4096 -M 2 /dev/sdb1

Why wrong: -M is invalid; use -m for reserved blocks percentage.

The multipathd daemon needs to be restarted

Why wrong: Restarting is unnecessary; multipathd is already running and would still show failure if LUN is not presented.

The /etc/multipath.conf needs to be edited to include the new LUN

Why wrong: Multipath discovers LUNs automatically; config editing is not required for detection.

The scsi device needs to be manually created with mknod

Why wrong: SCSI devices are created automatically by udev; manual mknod is not required.

lvextend -l +100%FREE /dev/vg/home; resize2fs /dev/vg/home

Why wrong: Incorrect. The `-l +100%FREE` option extends the logical volume by all free extents, but the question requires exactly 5 GB. Also, the order is correct but the size is wrong.

resize2fs /dev/vg/home; lvextend -L +5G /dev/vg/home

Why wrong: Incorrect. Reversing the order (filesystem resize before LV extend) will fail because the filesystem cannot be larger than the underlying block device.

lvextend -L +5G /dev/vg/home; mount -o remount /dev/vg/home

Why wrong: Incorrect. While `lvextend -L +5G` correctly extends the LV, `mount -o remount` does not resize the ext4 filesystem; it only remounts the filesystem without changing its size.

Increase the stripe size and recreate the LV

Why wrong: Increasing stripe size might help for large sequential I/O but requires recreation and does not solve the root issue of slow disks.

Add an additional PV to the VG and extend the LV

Why wrong: Adding more PVs may distribute load but does not address high I/O wait due to slow disks.

Convert the LV to use linear mapping instead of striping

Why wrong: Striping improves parallelism; converting to linear would worsen performance.

dumpe2fs /dev/sda1

Why wrong: dumpe2fs shows filesystem superblock and blocks group info, not for integrity checking.

fsck.ext4 -f /dev/sda1

Why wrong: The -f flag forces a check even if the filesystem is marked clean, which is not the best first step.

fsck -N /dev/sda1

Why wrong: The -N flag only shows what would be done without actually checking.

/dev/vg/data /data xfs defaults,noatime,swidth=2048 0 0

Why wrong: swidth sets the stripe width, not the stripe unit.

/dev/vg/data /data xfs defaults,noatime,allocsize=1m 0 0

Why wrong: allocsize controls space allocation chunk size, not stripe unit.

/dev/vg/data /data xfs defaults,noatime,sunit=1024 0 0

Why wrong: sunit=1024 gives a stripe unit of 512 KiB, not 1 MiB.