CompTIA Server+ Practice Test (SK0-005)

For maximum availability the two power supplies must follow different electrical paths all the way back to separate utility feeds. Plugging one PSU into the PDU that is fed by UPS-A and the other into the PDU fed by UPS-B creates two independent (2N) power paths. If Utility #1, UPS-A, or PDU-A fails, the server continues to draw power through Utility #2, UPS-B, and PDU-B, meeting the SLA. The other options leave both PSUs on the same power chain or leave one supply unused; a single failure in that chain would still drop the server, so they do not meet the requirement.

The correct answer explains that a SAS drive cannot function when connected to a SATA-only controller. While SAS controllers are backward-compatible with SATA drives, the reverse is not true. SATA controllers lack the necessary components and protocol support (the SCSI command set) to communicate with SAS drives.

• The option suggesting a different power connector is incorrect. While the connectors have slight physical differences, they are designed to be mechanically compatible in a way that allows SATA drives to connect to SAS backplanes, and the power delivery is standardized. The issue is electronic and protocol-based, not a power connector mismatch.
• The option regarding drive capacity is incorrect because the server already supports other drives of the same 4TB capacity, making a controller limit at that size highly improbable.
• The option about file system formatting is incorrect because drive detection by the server's firmware (BIOS/UEFI) and the operating system's storage controller is a prerequisite for formatting. A drive must be recognized as a hardware device before it can be partitioned and formatted.

Inside the single-bracket test command ([ ]) a numeric comparison must use the arithmetic primaries defined by POSIX. The operator -gt evaluates whether the first integer is greater than the second, matching the requirement that the size "exceeds" the limit. -ge would also allow the values to be equal, which is not requested. Escaping > still leaves a lexicographical string comparison, not a numeric one. The double-parentheses form (( SIZE > MAX )) is a Bash extension and is not guaranteed to work in a strictly POSIX /bin/sh script. Therefore, changing the line to if [ "$SIZE" -gt "$MAX" ]; then is the only portable fix.

A wired wrist strap must bond the technician to the same electrical potential as the equipment being serviced. Clipping the strap's lead to an unpainted, conductive part of the server chassis places the body and the memory sockets at the same potential, so any static charge on the technician is safely bled away through the strap's built-in 1 MΩ resistor. Doing this after the power cords are unplugged eliminates shock hazards while still using the chassis as the common point ground.

The other choices do not create a safe, reliable path:

• The outside of a shielding bag is deliberately conductive but is not referenced to the server, so charge can still jump to the DIMM.
• Powder-coated rack rails are insulated; the paint prevents an electrical connection.
• Connecting the strap to a facility UPS while the server remains on mains power bonds the technician to a different ground and violates ESD and electrical-safety best practice.

The message indicates that the dynamic linker cannot find libpq.so.5, a shared object supplied by the postgresql-libs package. Installing that package places the library in /usr/lib64 (and registers it with ldconfig), allowing the binary to load all required objects and start. Re-running ldconfig alone will not help when the library is completely absent. Firewall ports and kernel-module initramfs rebuilding are unrelated to user-space shared-library resolution, so those actions would not correct the failure.

The correct answer is Template deployment. This method involves creating a master or "golden" virtual machine with the operating system, applications, and all configurations already installed and set up. This template is then used to rapidly create multiple identical clones, which is the most efficient and consistent way to deploy a large number of identical VMs. An unattended installation automates the OS installation process but typically does not include pre-configured applications, making it less efficient for this scenario. A bare-metal installation is incorrect because it refers to installing an OS directly on physical hardware, not creating VMs. P2V conversion is the process of migrating an existing physical server to a virtual one, which is not applicable to deploying new VMs.

Safe-lifting guidance from OSHA and industry ergonomics references states that heavy objects should be kept as close to the body as possible, the spine should stay neutral, the knees and hips should be bent in a squat, and the weight should be raised smoothly with the legs. Twisting, lifting with a straight back, extending the load away from the torso, or raising it above shoulder height all increase the moment arm on the spine and greatly raise the chance of strains. Therefore the only option that follows accepted technique is the one that keeps the UPS close, maintains a straight back, and uses the legs for the lift.

The correct answer is a corrupt filesystem table. The combination of symptoms after a sudden power outage-including 'Input/output' errors on specific directories, missing files, and log entries about filesystem inconsistencies-points directly to logical corruption of the filesystem's metadata. A RAID cache battery failure is a potential underlying cause for the corruption if write-back caching was enabled, but the immediate problem causing the errors is the corruption itself. Multiple drive failure or loose connectors would more likely cause the entire logical volume to be inaccessible, not just produce errors on specific directories.

PowerShell performs variable expansion only in double-quoted strings. A double-quoted here-string (opened with @" and closed with "@ on a new line) behaves like a multi-line double-quoted string, so any $Variable inside it is replaced with the variable's value. A single-quoted here-string (@' ... '@) and a single-quoted string (' ... ') are literal, so \(ServerName would appear verbatim. Escaping the dollar sign with a backtick (`\)) also prevents expansion. Therefore, the only option that meets all three requirements-multi-line, exact formatting, and variable substitution-is the double-quoted here-string.

Running Sysprep with the /generalize switch removes computer-specific data-including the machine SID, event-log history, and activation identifiers-so that every cloned instance generates unique values during its first boot (OOBE). This procedure is Microsoft's supported method for preparing a Windows installation that will be captured or physically duplicated. Converting the disk to MBR, disabling Secure Boot, or replacing the hardware duplicator with MDT imaging does not address duplicate SID or activation issues and therefore will not prevent the cloning problems described.

The correct answer is the Service Level Agreement (SLA). An SLA is a formal contract between a service provider (like an IT department) and a customer (like a user department) that defines the level of service to be provided. It specifies measurable metrics such as uptime, availability, and response times. Consulting the SLA is the proper first step to determine if the application's current performance constitutes a breach of the agreed-upon standards.

• A Business Impact Analysis (BIA) is incorrect because it is used to identify critical business functions and the potential effects of their disruption. While the BIA informs the requirements of an SLA, it does not contain the specific performance metrics itself.
• A server configuration baseline is incorrect as it documents a system's standard, approved configuration at a point in time to manage changes and prevent configuration drift. It does not define service guarantees to end-users.
• A network infrastructure diagram is also incorrect. This document provides a visual map of the network's physical and logical layout, which is useful for troubleshooting connectivity but does not contain contractually obligated performance metrics.

The correct answer is CPU overheating. The key piece of evidence from the baseboard management controller (BMC) logs is the specific alert that the CPU temperature exceeded the shutdown threshold right before the system crashed. Modern servers and CPUs have built-in thermal protection that will automatically shut down the system to prevent permanent hardware damage from excessive heat.

A faulty memory module typically causes different symptoms, such as a Blue Screen of Death (BSOD), kernel panic, or specific memory-related errors in the logs, not a temperature warning. Corrupted system files are a software issue, not a hardware failure, and would likely result in boot errors or application-level problems rather than a hardware-level shutdown preceded by a temperature alert. A failing PSU would cause an abrupt loss of power but would not generate a CPU-specific temperature warning in the hardware logs before the event.

Only 12 Gb/s Serial Attached SCSI (SAS-3) drives meet every stated need. SAS-3 provides a per-lane throughput of 12 Gb/s and is full-duplex. Enterprise SAS drives include two independent ports, allowing each disk to be cabled to separate host bus adapters so the operating system can establish redundant, multipath access.

Serial ATA III is limited to 6 Gb/s, is half-duplex, and exposes only a single port per drive, so it cannot provide simultaneous redundant paths. USB 3.2 Gen 2 offers 10 Gb/s signaling but is designed for removable devices, is single-ported, and lacks the dual-path capability expected in enterprise backplanes. SDXC UHS-II media peaks around 312 MB/s (≈2.5 Gb/s), far below the required bandwidth and with no dual-port feature. Therefore, 12 Gb/s SAS-3 is the correct choice, while the other interfaces fail on speed, redundancy, or both.

RAID 10 (striped mirrors) has no parity overhead, so write I/O is limited only by the spindle count and controller cache; it still tolerates one disk failure per mirrored pair. RAID 50, RAID 6, and other parity-based levels must read old parity, recalculate it, and write new parity on every write, introducing significant latency-exactly the behaviour now seen on the controller. Simply changing stripe size or enabling read-ahead does not remove the parity penalty, and RAID 6 adds even more write overhead. Re-creating the volume as RAID 10 therefore resolves the performance issue while maintaining redundancy.

The correct answer is a clean agent suppression system. These systems use gaseous agents, such as HFC-227ea (FM-200) or inert gas mixtures (like Inergen), that extinguish fires without leaving residue, conducting electricity, or causing water damage to electronic components. They are also considered safe for use in occupied spaces.

A carbon dioxide (CO2) system is incorrect because, while it is a gas and leaves no residue, it extinguishes fires by displacing oxygen to levels that are lethal to humans, failing the personnel safety requirement.

A pre-action sprinkler system is a plausible but incorrect choice. While it is the safest type of water-based system for a data center because it requires a dual-trigger (e.g., smoke detection and sprinkler head activation) to release water, it ultimately still uses water, which would damage the electronic equipment.

A wet pipe sprinkler system is incorrect as it keeps water in the pipes at all times, posing a significant and constant risk of leaks and accidental discharge that could destroy server hardware.

The correct answer is that an external, certified vendor provides auditable proof of destruction and assumes liability. Financial services companies are subject to strict regulations like the Gramm-Leach-Bliley Act (GLBA), which mandates the protection of customer information. A certified vendor specializing in IT Asset Disposition (ITAD), such as one with R2 or e-Stewards certification, provides a Certificate of Destruction. This document serves as a legal and auditable record that data has been destroyed in a compliant manner, transferring the liability from the company to the vendor.

• Allowing the internal team to focus on core projects is a benefit, but it is secondary to the primary driver of compliance and liability mitigation.
• Repurposing components internally is an advantage of internal recycling, not external, and it may conflict with the stringent data destruction requirements for a financial firm.
• Using an external vendor typically incurs higher direct costs than handling disposal internally, making the claim of lower out-of-pocket costs incorrect.

Copying files with the -a (archive) option preserves all attributes, including SELinux labels. As a result, the PDFs carried the user_home_t type from the users' home directories into /var/www/html. By default the httpd_t domain is not allowed to read objects of type user_home_t, so SELinux blocks the access and the web server returns HTTP 403.

The supported fix is to relabel the affected path so that it regains the default httpd_sys_content_t type. The restorecon -R command consults the active policy and recursively applies the correct label to every item under /var/www/html, restoring normal access without weakening SELinux. Disabling SELinux (setenforce 0) would work but removes the protection entirely. Changing Apache directives or file ownership does not address the label mismatch, so access would still fail until the contexts are corrected.

The POST message indicates that the system firmware (BIOS/UEFI) lacks the microcode needed to initialize the newer-stepping CPUs. Updating the server's BIOS/UEFI to the latest revision supplied by the manufacturer loads the correct microcode and resolves the incompatibility. Adjusting voltages, replacing the power supply, or clearing the CMOS will not add the missing microcode, so those steps will not allow the system to boot with the new processors.

The correct action is to completely remove the power cables from the PDU to the server and then properly recycle them. According to the National Electrical Code (NEC), accessible portions of abandoned cables must be removed unless they are tagged for future use. Leaving abandoned cables, including power cables, in place can create safety hazards, impede airflow needed for cooling active equipment, and complicate future cable management. This removal process is often referred to as cable remediation or cable mining.

Leaving cables coiled in the rack, even if neat, contributes to a 'spaghetti' environment, blocks critical airflow, and can be mistaken for live cables. Cutting the ends off the cables is dangerous as the other end may still be connected to a live PDU, creating a serious electrical hazard. Tagging cables as 'abandoned' while leaving them connected to the PDU does not resolve the physical obstruction or potential safety issues and goes against best practices, which favor complete removal.

MBR-partitioned disks that boot from legacy BIOS firmware are limited to addressing about 2 TB. Converting the disk to the GUID Partition Table (GPT) removes that size ceiling, but Windows can only boot from a GPT disk when the server firmware is set to UEFI mode. Switching the firmware to UEFI and initializing the disk as GPT therefore lets Windows Server use and boot from the entire 4 TB drive.

ReFS formatting, dynamic disks, or NTFS compression do not bypass the 2 TB barrier imposed by an MBR boot disk, so they would still leave part of the drive inaccessible or unbootable.