NOTE: This question is referring to a system which is Logical/Technical, so it is in the context of a system that you must choose the right answer. This is very important to read the question carefully and to identify the context whether it is in the Physical world or in the Technical/Logical world.
RFC 2828 (Internet Security Glossary) defines fail safe as a mode of system termination that automatically leaves system processes and components in a secure state when a failure occurs or is detected in the system.
A secure state means in the Logical/Technical world that no access would be granted or no packets would be allowed to flow through the system inspecting the packets such as a firewall for example.
If the question would have made reference to a building or something specific to the Physical world then the answer would have been different. In the Physical World everything becomes open and full access would be granted. See the valid choices below for the
Physical context.
Fail-safe in the physical security world is when doors are unlocked automatically in case of
emergency. Used in environment where humans work around. As human safety is prime
concern during Fire or other hazards.
The following were all wrong choices:
Fail-secure in the physical security world is when doors are locked automatically in case of
emergency. Can be in an area like Cash Locker Room provided there should be alternative
manually operated exit door in case of emergency.
Fail soft is selective termination of affected non-essential system functions and processes
when a failure occurs or is detected in the system.
Fail Over is a redundancy mechanism and does not apply to this question.
There is a great post within the CCCure Forums on this specific questions:
saintrockz who is a long term contributor to the forums did outstanding research and you
have the results below. The CCCure forum is a gold mine where thousands of questions related to the CBK have been discussed.
According to the Official ISC2 Study Guide (OIG):
Fault Tolerance is defined as built-in capability of a system to provide continued correct
execution in the presence of a limited number of hardware or software faults. It means a
system can operate in the presence of hardware component failures. A single component
failure in a fault-tolerant system will not cause a system interruption because the alternate
component will take over the task transparently. As the cost of components continues to
drop, and the demand for system availability increases, many non-fault-tolerant systems
have redundancy built-in at the subsystem level. As a result, many non-fault-tolerant
systems can tolerate hardware faults - consequently, the line between a fault-tolerant
system and a non-fault-tolerant system becomes increasingly blurred.
According to Common Criteria:
Fail Secure - Failure with preservation of secure state, which requires that the TSF (TOE
security functions) preserve a secure state in the face of the identified failures.
Acc. to The CISSP Prep Guide, Gold Ed.:
Fail over - When one system/application fails, operations will automatically switch to the
backup system.
Fail safe - Pertaining to the automatic protection of programs and/or processing systems to
maintain safety when a hardware or software failure is detected in a system.
Fail secure - The system preserves a secure state during and after identified failures occur.
Fail soft - Pertaining to the selective termination of affected non-essential processing when
a hardware or software failure is detected in a system.
Acc. to CISSP for Dummies:
Fail closed - A control failure that results all accesses blocked.
Fail open - A control failure that results in all accesses permitted.
Failover - A failure mode where, if a hardware or software failure is detected, the system
automatically transfers processing to a hot backup component, such as a clustered server.
Fail-safe - A failure mode where, if a hardware or software failure is detected, program
execution is terminated, and the system is protected from compromise.
Fail-soft (or resilient) - A failure mode where, if a hardware or software failure is detected,
certain, noncritical processing is terminated, and the computer or network continues to
function in a degraded mode.
Fault-tolerant - A system that continues to operate following failure of a computer or
network component.
It's good to differentiate this concept in Physical Security as well:
Fail-safe
Door defaults to being unlocked
Dictated by fire codes
Door defaults to being locked
Fail-secure
Reference(s) used for this question:
SHIREY, Robert W., RFC2828: Internet Security Glossary, may 2000.

Administrative, technical and physical controls are categories of access control mechanisms.
Logical and Technical controls are synonymous. So both of them could be eliminated as possible choices.
Physical Controls: These are controls to protect the organization's people and physical environment, such as locks, gates, and guards. Physical controls may be called "operational controls" in some contexts.
Physical security covers a broad spectrum of controls to protect the physical assets (primarily the people) in an organization. Physical Controls are sometimes referred to as "operational" controls in some risk management frameworks. These controls range from doors, locks, and windows to environment controls, construction standards, and guards. Typically, physical security is based on the notion of establishing security zones or concentric areas within a facility that require increased security as you get closer to the valuable assets inside the facility. Security zones are the physical representation of the defense-in-depth principle discussed earlier in this chapter. Typically, security zones are associated with rooms, offices, floors, or smaller elements, such as a cabinet or storage locker. The design of the physical security controls within the facility must take into account the protection of the asset as well as the individuals working in that area.
Reference(s) used for this question:
Hernandez CISSP, Steven (2012-12-21). Official (ISC)2 Guide to the CISSP CBK, Third Edition ((ISC)2 Press) (Kindle Locations 1301-1303). Auerbach Publications. Kindle Edition. and Hernandez CISSP, Steven (2012-12-21). Official (ISC)2 Guide to the CISSP CBK, Third Edition ((ISC)2 Press) (Kindle Locations 1312-1318). Auerbach Publications. Kindle Edition.

Section: Network and Telecommunications
Explanation/Reference:
Before a LAN is connected to the Internet, you need to determine what the access controls mechanisms are to be used, this would include how you are going to authenticate individuals that may access your network externally through access control.
The following answers are incorrect:
plan for implementing workstation locking mechanisms. This is incorrect because locking the workstations have no impact on the LAN or Internet access.
plan for protecting the modem pool. This is incorrect because protecting the modem pool has no impact on the LAN or Internet access, it just protects the modem.
plan for providing the user with his account usage information. This is incorrect because the question asks what should be done first. While important your primary concern should be focused on security.

All of these statements pertaining to SSL are true except that it is primary
use is to authenticate the client to the server using public key cryptography and digital
certificates. It is the opposite, Its primary use is to authenticate the server to the client.
The following reference(s) were used for this question:
KRUTZ, Ronald L. & VINES, Russel D., The CISSP Prep Guide: Mastering the Ten
Domains of Computer Security, John Wiley & Sons, 2001, Chapter 4: Cryptography (page
170).

Explanation/Reference:
To apply this concept to access control, the pair of elements is the subject and object, and the subject has to have an upper bound equal or higher than the object being accessed.
WIKIPEDIA has a great explanation as well:
In computer security, lattice-based access control (LBAC) is a complex access control based on the interaction between any combination of objects (such as resources, computers, and applications) and subjects (such as individuals, groups or organizations).
In this type of label-based mandatory access control model, a lattice is used to define the levels of security that an object may have and that a subject may have access to. The subject is only allowed to access an object if the security level of the subject is greater than or equal to that of the object.
Reference(s) used for this question:
KRUTZ, Ronald L. & VINES, Russel D., The CISSP Prep Guide: Mastering the Ten Domains of Computer Security, 2001, John Wiley & Sons, Page 34.
and
http://en.wikipedia.org/wiki/Lattice-based_access_control