No sign-up needed

Free CompTIA Server+ practice test

20 questions, each with a clear explanation.

Quick PracticeQuestion 1 of 20
Your progress through this set
Server Hardware Installation and ManagementStep 1 · Choose one answer

A team is filling an empty 42U rack with two 4U storage arrays, which are by far the heaviest units, and eight 1U servers. The rack's leveling pads are lowered and its stabilizer brackets are fitted. Where should the two storage arrays go?

Question review

All 20 CompTIA Server+ questions with explanations

Try the set above first, then use this list to review. Each answer and explanation stays closed until you open it.

  1. Question 1 · Server Hardware Installation and Management

    A team is filling an empty 42U rack with two 4U storage arrays, which are by far the heaviest units, and eight 1U servers. The rack's leveling pads are lowered and its stabilizer brackets are fitted. Where should the two storage arrays go?

    1. At the top of the rack, so the heat they give off can rise out of it
    2. At the bottom of the rack, installed first, before the lighter servers
    3. In the middle of the rack, so their weight sits between the servers
    4. Wherever there is space, since the stabilizer brackets stop tipping
    Show answer and explanation for question 1

    Correct answer: B · At the bottom of the rack, installed first, before the lighter servers

    Lenovo's rack safety statements tell installers to lower the leveling pads, fit stabilizer brackets, always install the heaviest devices in the bottom of the rack cabinet, and install devices starting from the bottom. Putting the two arrays in first, at the bottom, keeps the rack's center of gravity low. Mounting them at the top raises the center of gravity and makes tipping more likely, whatever the airflow argument. The middle of the rack still leaves lighter servers underneath the heaviest load. Stabilizer brackets are one precaution in the same list, so they add to the bottom-first rule rather than replacing it.

    Look for this next time: Load a rack from the bottom up, with the heaviest equipment lowest.

  2. Question 2 · Server Hardware Installation and Management

    A company is building a two-host virtualization cluster. Both hosts must see the same shared volumes as SCSI disks, and all storage traffic must use the existing 10 GbE Ethernet switches, because there is no budget for Fibre Channel HBAs or switches. Which shared storage option fits?

    1. An iSCSI SAN, which carries SCSI commands to the shared disks over TCP/IP
    2. An NFS export from a NAS, which shares files with both hosts over the LAN
    3. A Fibre Channel SAN, using new HBAs and new FC switches throughout
    4. A JBOD shelf cabled to each host, with one separate shelf for each host
    Show answer and explanation for question 2

    Correct answer: A · An iSCSI SAN, which carries SCSI commands to the shared disks over TCP/IP

    RFC 7143 describes iSCSI as a transport protocol for SCSI that works on top of TCP, so a storage array can present logical units to hosts across an ordinary Ethernet and IP network, and each host sees the shared volume as a SCSI disk. NFS, which RFC 7530 describes as a distributed file system protocol, gives hosts access to files on a server rather than a shared SCSI disk. A Fibre Channel SAN could provide shared disks, but it needs the HBAs and FC switches the budget rules out. A separate JBOD shelf for each host gives every host its own disks, so nothing is shared.

    Look for this next time: iSCSI delivers shared SCSI disks over an existing Ethernet and IP network.

  3. Question 3 · Server Hardware Installation and Management

    A server in a remote data center has stopped responding: its operating system doesn't answer on the production network, and nobody is on site. The server's management controller, a Dell iDRAC, is cabled to a separate management network. How can the admin see the console and restart the server?

    1. Open an RDP session to the server's production IP address and restart it
    2. Use the iDRAC on the management network for its console and power control
    3. Connect over SSH to the production IP address and run a reboot command
    4. Ask the network team to shut and reopen the server's switch port
    Show answer and explanation for question 3

    Correct answer: B · Use the iDRAC on the management network for its console and power control

    Dell's iDRAC user's guide says the controller helps administrators manage servers remotely and reduces the need for physical access, regardless of operating system or hypervisor presence or state. Its remote presence features include power control and, with the right license, a virtual console, and rack and tower servers can give it a dedicated network adapter. That out-of-band path works while the operating system is frozen. RDP and SSH both depend on the frozen operating system answering on the production network. Shutting and reopening the switch port only resets the network link, so it neither restarts the server nor shows its console.

    Look for this next time: When the OS is unreachable, use the out-of-band management controller for console and power.

  4. Question 4 · Server Administration

    An admin is installing a 64-bit edition of Windows Server on a new server whose firmware runs in UEFI mode, and the admin wants to keep UEFI. Setup won't install to the system disk, which an earlier test left with an MBR partition table. The disk holds no data. What should the admin do?

    1. Create an extended partition on the MBR disk for the system volume
    2. Convert the MBR disk to a dynamic disk before running Setup again
    3. Switch the server's firmware to legacy BIOS mode for this install
    4. Clear the disk and initialize it with a GUID partition table (GPT)
    Show answer and explanation for question 4

    Correct answer: D · Clear the disk and initialize it with a GUID partition table (GPT)

    Microsoft's Windows and GPT FAQ says 64-bit editions from Windows Server 2008 on can boot from GPT disks on UEFI-based systems, that UEFI systems require the boot partition to reside on a GPT disk, and that Windows does not support booting UEFI from MBR disks. Because the disk is empty, clearing it and initializing it as GPT fixes the problem and keeps UEFI. An extended partition is the MBR workaround for more than four partitions, so the disk stays MBR. A dynamic disk can use either partition style, so converting changes nothing here. Legacy BIOS mode might install, but it drops the UEFI requirement in the stem.

    Look for this next time: Windows boots in UEFI mode only from a GPT disk.

  5. Question 5 · Server Administration

    A new file server will be reachable only inside the company network. The network team says its IPv4 address must come from the private address space defined in RFC 1918. Which address meets that rule?

    1. 172.32.40.10
    2. 100.64.40.10
    3. 172.31.40.10
    4. 192.169.40.10
    Show answer and explanation for question 5

    Correct answer: C · 172.31.40.10

    RFC 1918 reserves three blocks for private internets: 10.0.0.0 to 10.255.255.255, 172.16.0.0 to 172.31.255.255, and 192.168.0.0 to 192.168.255.255. The address 172.31.40.10 falls inside the 172.16/12 block. 172.32.40.10 sits just past the end of that block, so it is outside every RFC 1918 range. 100.64.40.10 belongs to 100.64.0.0/10, which RFC 6598 reserves as Shared Address Space for carrier-grade NAT and describes as distinct from RFC 1918 private space. 192.169.40.10 is outside 192.168.0.0/16, because the second octet would have to be 168.

    Look for this next time: RFC 1918 private space is 10/8, 172.16/12 (172.16 to 172.31), and 192.168/16.

  6. Question 6 · Server Administration

    Two database servers share storage. During normal operations, Node A runs the database for all users while Node B runs no workload. Node B watches Node A through heartbeat signals on a private cluster network and takes over the database if Node A fails. Which configuration is this?

    1. An active-passive failover cluster with one node on standby
    2. An active-active cluster that spreads load across both nodes
    3. Round-robin load balancing that alternates between the nodes
    4. NIC teaming that pairs two network adapters in one server
    Show answer and explanation for question 6

    Correct answer: A · An active-passive failover cluster with one node on standby

    Microsoft's Failover Clustering documentation describes an active-passive cluster as one where some nodes are active while others stay on standby, ready to take over if an active node fails, and says clusters watch node health through heartbeat signals and can fail over automatically. Node B's idle, watching role matches that description. In Microsoft's description of an active-active cluster, all nodes work at the same time to balance the workload, which is not happening here. Round-robin load balancing hands requests to each server in turn. NIC teaming, which CompTIA lists under redundant server networking, pairs adapters within a single server.

    Look for this next time: One node serving while a standby node waits to take over is an active-passive cluster.

  7. Question 7 · Server Administration

    A test web server runs as a VM in Oracle VM VirtualBox, and its network adapter is in NAT mode. Coworkers on the office LAN must browse to it at its own IP address on that LAN, and right now they can't reach it at all. Which networking mode should the admin set for the VM's adapter?

    1. Internal networking, which links the VM only to other chosen VMs
    2. Host-only networking, which connects the VM to the host and VMs
    3. Bridged networking, which puts the VM on the physical network
    4. NAT without port forwarding, which keeps the current setting
    Show answer and explanation for question 7

    Correct answer: C · Bridged networking, which puts the VM on the physical network

    The VirtualBox manual says a VM in NAT mode sits behind a router built into VirtualBox, is unreachable from outside, and cannot run a server that way unless port forwarding is set up. It describes bridged networking as the mode for running servers in a guest: VirtualBox exchanges packets directly with the physical network adapter, so the guest looks as if it were cabled to the network and can hold its own LAN address. Internal networking is visible only to selected virtual machines, not to the host or the outside world. Host-only networking connects the host and its VMs without using the physical network adapter.

    Look for this next time: A VM that must be reached at its own LAN address needs bridged networking.

  8. Question 8 · Server Administration

    An admin's Bash script stores the names of 25 web servers in an array named servers. The script must run the same restart command against every name in that array without the command appearing 25 times in the code. Which construct does that?

    1. An if statement that runs the restart when its test succeeds
    2. A case statement that matches one name to one block of commands
    3. A comment line above the command that lists all 25 names
    4. A for loop that runs the restart once for each array element
    Show answer and explanation for question 8

    Correct answer: D · A for loop that runs the restart once for each array element

    The GNU Bash Reference Manual says the for command expands a list of words and then runs its commands once for each word, with a variable set to the current word, and it documents indexed arrays as one way to hold such a list. Looping over the array's elements runs the restart once per server name. An if statement runs its commands once, when its test command returns a zero status, so it does not repeat them across a list. A case statement runs the one block whose pattern matches a word. Bash ignores a comment when the script runs, so a list in a comment does nothing.

    Look for this next time: Use a loop to repeat the same command for each item in a list or array.

  9. Question 9 · Server Administration

    After a weekend of undocumented fixes, a file server starts behaving oddly. The team keeps a formally approved set of specifications for this server, agreed at a set point in time, that can be changed only through change control. What is that document, and how does it help now?

    1. The baseline configuration; compare the live server against it
    2. The service level agreement; check its promised uptime
    3. The asset inventory; confirm the make, model, and serial number
    4. The MTTR report; look up how long past repairs usually took
    Show answer and explanation for question 9

    Correct answer: A · The baseline configuration; compare the live server against it

    NIST's glossary, citing SP 800-128, defines a baseline configuration as a set of specifications for a system that has been formally reviewed and agreed on at a given point in time and can be changed only through change control procedures, and it is used as the basis for future builds, releases, and changes. Comparing the live server against that baseline shows what the weekend fixes changed. A service level agreement states service commitments, such as uptime, rather than the server's settings. An asset inventory identifies the hardware, not its configuration. A mean time to recover figure describes repair speed and says nothing about what changed.

    Look for this next time: An approved baseline configuration is the reference for spotting unauthorized changes.

  10. Question 10 · Security and Disaster Recovery

    Someone added an unknown account to a Windows file server's local Administrators group, and no record shows who did it. The security team wants every future change to that group's membership logged, including the account that made the change. Which audit setting should be enabled?

    1. Audit Logon, so the server records each attempt to sign in to it
    2. Audit File System, so the server records access to shared files
    3. An account lockout threshold for failed sign-ins
    4. Audit Security Group Management, with Success events enabled
    Show answer and explanation for question 10

    Correct answer: D · Audit Security Group Management, with Success events enabled

    Microsoft's documentation says the Audit Security Group Management subcategory determines whether Windows generates audit events when a member is added to or removed from a security group, and it recommends Success auditing. Event 4732 records each member added to a security-enabled local group, such as the built-in Administrators group, and its Subject fields name the account that requested the change. Audit Logon covers attempts to log on to a computer, not group membership. Audit File System covers attempts to access file system objects. An account lockout threshold limits password guessing and records nothing about who changed a group.

    Look for this next time: Auditing security group management records who added whom to privileged groups.

  11. Question 11 · Security and Disaster Recovery

    A new server will run only an internal web application. A review finds that the default install also left Telnet, an FTP service, and file and print sharing running, and the application needs none of them. Which step best reduces the server's attack surface?

    1. Keep the services running but move each one to a nonstandard port
    2. Keep the services and require longer passwords for Telnet and FTP
    3. Remove the unneeded services, and disable any that can't be removed
    4. Install antivirus software and leave the services as installed
    Show answer and explanation for question 11

    Correct answer: C · Remove the unneeded services, and disable any that can't be removed

    NIST SP 800-123 tells administrators to remove all services, applications, and network protocols that are not required and to disable any unnecessary components that cannot be removed. It explains that each service added to a host increases the risk of compromise, because each one is another possible avenue of access for an attacker. Removing Telnet, FTP, and file and print sharing closes those avenues on a single-purpose web server. Moving the services to other ports leaves them running and reachable. Longer passwords still leave Telnet's unencrypted remote access in place. Antivirus software adds host protection but keeps every unneeded service exposed.

    Look for this next time: Harden a server by removing the services it doesn't need, and disabling what can't be removed.

  12. Question 12 · Security and Disaster Recovery

    A company is retiring storage from several servers. A bin of failed drives, both hard disk drives and SSDs, held sensitive customer data, and none of the drives powers on anymore. The recycler offers to degauss every drive in the bin. What should happen to the SSDs?

    1. Degauss them with the hard disk drives, since one pass covers both
    2. Quick-format them in a working server before they go to recycling
    3. Run each drive's built-in sanitize command first
    4. Physically destroy them, for example by shredding or disintegration
    Show answer and explanation for question 12

    Correct answer: D · Physically destroy them, for example by shredding or disintegration

    NIST SP 800-88 Rev. 2 says degaussing should not be used for non-magnetic media such as flash storage and SSDs, and notes that degaussing an SSD can complete successfully while no sensitive data is sanitized. It also says destructive techniques may be the only option when a storage device fails, and it lists shredding, disintegration, pulverizing, melting, and incineration as destroy techniques. So degaussing the SSDs with the hard disk drives would leave their data in place. A quick format and a built-in sanitize command both need a drive that powers on, and these drives don't.

    Look for this next time: Degaussing doesn't sanitize SSDs; failed flash drives with sensitive data are destroyed.

  13. Question 13 · Security and Disaster Recovery

    Three days ago, a faulty script deleted 200 customer records from a production SQL Server database. Since then, the database has taken thousands of valid new orders. The admin has a full backup from just before the deletion. How can the admin recover the 200 records without losing the new orders?

    1. Restore the backup as a separate copy, then copy the records back
    2. Restore the backup over the production database in a quiet hour
    3. Take a fresh full backup of production now and restore that one
    4. Restore the backup on a second server and move users over to it
    Show answer and explanation for question 13

    Correct answer: A · Restore the backup as a separate copy, then copy the records back

    Microsoft's SQL Server documentation explains how to restore a database to a new location and optionally rename it, which creates a copy of the database on the same server instance or a different one. Restoring the old backup side by side under a new name leaves production untouched, and the admin can then copy the 200 deleted records back. Restoring over production would replace it with the three-day-old copy and lose the new orders. A fresh backup taken now no longer contains the deleted records. Moving users to a restored server puts them on the old data and drops three days of orders.

    Look for this next time: Restore to an alternate location when you need old data without overwriting current data.

  14. Question 14 · Security and Disaster Recovery

    A company replicates its file server volumes to a second data center about 1,200 miles away over a high-latency link. With synchronous replication, every write waits on the remote site, and users complain that saving files is slow. The business accepts losing a few minutes of changes if the main site is lost. Which approach fits?

    1. Mirror each volume to a second disk inside the same file server
    2. Asynchronous replication, which fits long links and higher latency
    3. Synchronous replication over an added, equally slow link
    4. Nightly full backups copied to the second data center by the morning
    Show answer and explanation for question 14

    Correct answer: B · Asynchronous replication, which fits long links and higher latency

    Microsoft's Storage Replica documentation says synchronous replication mirrors data within a low-latency network site to ensure zero data loss, while asynchronous replication mirrors data across sites beyond metropolitan ranges over higher-latency links, without a guarantee that both sites hold identical copies at the moment of a failure. A 1,200-mile, high-latency link and a business that accepts a few minutes of loss fit the asynchronous description. An added link that is just as slow keeps the synchronous wait in place. Mirroring inside the same server offers no protection if the site is lost. Nightly backups could lose most of a day's changes.

    Look for this next time: Over long, high-latency links, asynchronous replication trades a small data-loss window for speed.

  15. Question 15 · Troubleshooting

    Users report that a file server drops connections several times an hour. The admin's theory was a faulty NIC driver update, so the admin rolled the driver back to the previous version as the fix, but the drops continue. Following CompTIA's troubleshooting methodology for CompTIA Server+, what should the admin do next?

    1. Keep the old driver and change the cable, port speed, and teaming at once
    2. Record the rollback as the fix and close the ticket, since it was a change
    3. Undo the rollback if appropriate, then try one new change and test it
    4. Stop testing theories and replace the server with a spare from stock
    Show answer and explanation for question 15

    Correct answer: C · Undo the rollback if appropriate, then try one new change and test it

    The SK0-005 objectives list CompTIA's troubleshooting steps, and the step for implementing a solution says to make one change at a time and confirm whether it resolved the problem, and, if it did not, to reverse the change, if appropriate, and implement a new change. Undoing the unsuccessful rollback and testing one new change follows that step. Changing the cable, port speed, and teaming together breaks the one-change rule, so any fix couldn't be traced to its cause. Closing the ticket skips verifying full system functionality while the drops continue. Swapping the whole server abandons the method without a tested theory.

    Look for this next time: Make one change at a time, and reverse a change that didn't fix the problem.

  16. Question 16 · Troubleshooting

    A database server's logs show a growing number of corrected ECC memory errors over the past two weeks, all traced to the DIMM in slot B2. The server hasn't crashed, and no data errors have been found. What should the admin do?

    1. Nothing yet, because ECC has corrected every one of the logged errors
    2. Schedule a maintenance window and replace the DIMM in slot B2
    3. Reinstall the operating system to clear the error history
    4. Turn off ECC in the firmware so the error messages stop appearing
    Show answer and explanation for question 16

    Correct answer: B · Schedule a maintenance window and replace the DIMM in slot B2

    The Linux kernel's RAS documentation explains that a corrected error is one the error-correction mechanism detected and fixed, so the system keeps running with no data damaged yet. It adds that proactive replacement of memory modules showing corrected errors can reduce the likelihood of uncorrected errors and system panics. A rising count on one DIMM is an early warning, so replacing that module in a planned window is the safe response. Doing nothing ignores the warning. Reinstalling the operating system leaves the failing module in place. Turning off ECC removes the protection that has been correcting the errors.

    Look for this next time: Rising corrected memory errors on one DIMM call for planned, proactive replacement.

  17. Question 17 · Troubleshooting

    After a power test, write performance on a server's RAID 10 volume drops sharply. The RAID controller's log reports a failed cache battery, and the volume's current write policy shows Write-Through, although it is configured for Write-Back. What is the most likely cause?

    1. A drive in one mirror failed, and the array is rebuilding to a spare
    2. The file system is fragmented and needs a defrag
    3. The controller fell back to write-through after the battery failed
    4. The operating system turned off disk caching after the power test
    Show answer and explanation for question 17

    Correct answer: C · The controller fell back to write-through after the battery failed

    Dell's PERC H700 and H800 user's guide says write-back caching is used whenever the battery is present and in good condition, and write-through caching is used whenever the battery is missing or in a low-charge state. In write-through mode the controller reports a write as complete only after the disks have received the data, rather than when its cache has, so writes slow down. A failed battery and a policy that now shows Write-Through point to that fallback, and replacing the battery restores write-back caching. The log reports no failed drive or rebuild. Fragmentation can't change the controller's write policy, and Dell's PERC documentation says the controller manages cached data independently of the operating system.

    Look for this next time: A failed RAID cache battery forces write-through mode and slower writes.

  18. Question 18 · Troubleshooting

    After patching, the app.service unit on a Linux server won't start. Its unit file contains Requires=db.service and After=db.service. Running systemctl status shows that db.service failed to start because its data directory moved during the patch. What should the admin fix first?

    1. Reinstall the application package that provides app.service
    2. Delete the Requires= line so app.service starts without db.service
    3. Reboot the server until app.service manages to start on its own
    4. Get db.service starting again, then start app.service after it
    Show answer and explanation for question 18

    Correct answer: D · Get db.service starting again, then start app.service after it

    The systemd.unit documentation says that when a unit lists another unit in Requires=, activating it activates the listed unit too, and if that unit fails to activate while an After= ordering dependency on it is set, the requiring unit will not be started. app.service is waiting on a failed db.service, so the fix starts with the database service's moved data directory. Reinstalling the application leaves the failed dependency in place. Deleting Requires= would let the application start without the database it needs. Rebooting repeats the same failed start, because db.service still points at the old directory.

    Look for this next time: When a service won't start, check whether a service it requires has failed first.

  19. Question 19 · Troubleshooting

    Last week, app01.corp.example moved from 10.0.5.20 to 10.0.8.20. Other servers now reach it at the new address, and dig app01.corp.example returns 10.0.8.20, yet one Linux admin server still connects to 10.0.5.20. Its /etc/nsswitch.conf contains the line hosts: files dns. What is the most likely cause?

    1. The DNS server's A record for app01 still points to the old address
    2. The admin server's /etc/hosts file still maps app01 to the old address
    3. The admin server's default gateway points at the wrong router
    4. The app01 server's host firewall is blocking ICMP from the admin server
    Show answer and explanation for question 19

    Correct answer: B · The admin server's /etc/hosts file still maps app01 to the old address

    The nsswitch.conf manual page says the order of the services on a database line sets the order in which they are queried until a result is found, and that the files source reads /etc/hosts for host lookups. With hosts: files dns, a stale /etc/hosts entry answers before DNS is ever asked. BIND's documentation describes dig as a tool that queries DNS name servers and displays the answers they return, so its correct result shows the DNS record was already updated. A wrong default gateway would affect reaching other networks, not the address a name resolves to. An ICMP block doesn't change name resolution either.

    Look for this next time: When dig and the system disagree, check local sources such as /etc/hosts that are consulted first.

  20. Question 20 · Troubleshooting

    On a Linux web server with SELinux enforcing, an admin uses mv to move a new page from a home directory into /var/www/html. The web server returns 403 Forbidden for that page only. Its permissions match the other pages, the audit log shows an SELinux denial for the web server reading it, and ls -Z shows it still has a home-directory label. What should the admin do?

    1. Run restorecon on the file to apply the directory's default label
    2. Set the file's permissions to 777 so any process can read it
    3. Put SELinux in permissive mode so the web server can read the page
    4. Restart the web server so it rereads /var/www/html
    Show answer and explanation for question 20

    Correct answer: A · Run restorecon on the file to apply the directory's default label

    The GNU coreutils manual notes that mv tries to keep extended attributes, which can include the SELinux context, and offers a -Z option that sets the system default context instead, much like restorecon. The restorecon manual page describes the tool as restoring files' default SELinux security contexts. A moved file that keeps its home-directory label is denied even though its permissions are fine, so restoring the default label for its new location clears the denial. Permissions of 777 change the wrong control and weaken security. The selinux manual page says permissive mode permits accesses the policy would deny, which removes protection instead of fixing the label. A restart leaves the label unchanged.

    Look for this next time: A file moved into place can keep its old SELinux label; restorecon applies the right one.

Use the sample well

The rationale matters more than the score.

A short set can show you what deserves another look. A strong run shows you can read a server symptom, a log entry, or a requirement and pick the step a server administrator would take next. Twenty questions are a small sample, so the result does not predict a 750 on the 100–900 scale.

01

Answer before reviewing

Choose the one answer that meets every condition in the scenario. Then read why each other option falls short.

02

Read every rationale

Confirm why the best option fits, even when you were right.

03

Follow the pattern

Use recurring misses—not one score—to choose your next focus.