Explore / AWS Certified Security - Specialty (SCS-C03) Exam Prep / Practice questions
AWS Certified Security - Specialty (SCS-C03) practice questions
10 questions from the bank behind this course's timed mock. Free to read and to answer, no account needed, and each has its answer and an explanation. They belong to AWS Certified Security - Specialty (SCS-C03) Exam Prep.
Answer them one at a time →Read lesson 1 free
Question 1
A security engineer is moving a containerised workload from self-managed Amazon EC2 instances to Amazon ECS on AWS Fargate. The team lead wants to know which security task the team will STILL own after the move under the AWS shared responsibility model. Which task is it?
- Patching the operating system of the underlying container host
- Maintaining the physical hardware that the Fargate tasks run on
- Applying guest OS patches to the EC2 instances that run the tasks
- Patching vulnerable packages in the container image the team builds
Show the answer
Answer: D. Patching vulnerable packages in the container image the team builds
On Fargate, AWS patches the host, but the customer still owns the container image and the packages inside it. Host OS patching was the customer's job on EC2 (guest OS) and becomes AWS's job on Fargate, so the team no longer does it.
Question 2
A security engineer must be alerted within minutes whenever the root user signs in to any account in the organization. An organization trail already delivers management events to an S3 bucket in the log archive account. Which approach meets the requirement with the least added effort?
- Create a second trail in each member account that records management events to its own S3 bucket
- Add advanced event selectors for S3 data events on the log archive bucket so every object write produces an alert
- Run aws cloudtrail validate-logs every hour against the trail and alert when it reports a root sign-in
- Configure the organization trail to also send events to CloudWatch Logs, then create a metric filter for root sign-in events and a CloudWatch alarm on that metric
Show the answer
Answer: D. Configure the organization trail to also send events to CloudWatch Logs, then create a metric filter for root sign-in events and a CloudWatch alarm on that metric
The lesson's pattern for near-real-time alerting is to send the trail to CloudWatch Logs and add a metric filter with an alarm. validate-logs only checks log file integrity through the digest files and does not look at what the events contain, so it cannot detect a sign-in.
Question 3
A company has subscribed to AWS Shield Advanced for its public Application Load Balancers. During a DDoS attack, the security team wants the Shield Response Team (SRT) to be able to update its AWS WAF rules. It also wants the SRT to call the SOC as soon as a protected resource's health degrades. Which TWO actions should the team take before any attack occurs? (Select TWO.)
More than one answer is right.
- Associate health checks with the protected resources and register SOC emergency contacts by using aws shield associate-proactive-engagement-details.
- Create an IAM role that trusts drt.shield.amazonaws.com, attach the AWSShieldDRTAccessPolicy managed policy, and associate it by using aws shield associate-drt-role.
- Create named break-glass IAM users for SRT engineers in the dedicated security account, with MFA and permission to assume the incident response role.
- Wait until an attack is detected, then grant the SRT temporary access by attaching AWS WAF permissions to the on-call engineer's existing IAM user.
- Make the Security Tooling account the delegated administrator for Amazon GuardDuty so that the SRT can view DDoS findings across the organization.
Show the answer
Answers: A. Associate health checks with the protected resources and register SOC emergency contacts by using aws shield associate-proactive-engagement-details.; B. Create an IAM role that trusts drt.shield.amazonaws.com, attach the AWSShieldDRTAccessPolicy managed policy, and associate it by using aws shield associate-drt-role.
The SRT gets access through a role that trusts the Shield DRT service principal, and proactive engagement needs both health checks on the protected resources and emergency contacts. Break-glass IAM users are for your own responders when federation fails, not for the AWS SRT, and the SRT works through its service role.
Question 4
A security engineer is containing a compromised EC2 instance and must keep it running for memory forensics. She creates a new security group, adds no inbound rules, and runs `aws ec2 modify-instance-attribute --groups` to make it the instance's only security group. Minutes later, VPC Flow Logs show the instance opening new outbound connections to a command-and-control service that keeps changing its IP address. Which action BEST fixes the containment gap without affecting other instances in the subnet?
- Add an outbound NACL deny rule for 0.0.0.0/0 to the instance's subnet
- Revoke the new security group's default egress rule that allows all outbound traffic to 0.0.0.0/0
- Stop the instance so it can no longer start new outbound connections
- Add an outbound deny rule for 0.0.0.0/0 to the isolation security group
Show the answer
Answer: B. Revoke the new security group's default egress rule that allows all outbound traffic to 0.0.0.0/0
A new security group starts with no inbound rules but does allow all outbound traffic, so that egress rule must be revoked before the group can isolate anything. A subnet-wide NACL deny would also stop the traffic, but it would cut off every healthy instance in the subnet too. Security groups have no deny rules, and stopping the instance wipes the memory evidence.
Question 5
A team runs a service on Amazon ECS with AWS Fargate. The task definition pulls a database password from AWS Secrets Manager and injects it into the container as an environment variable. Tasks now fail before the container starts, with an error saying the secret could not be retrieved. The application code never calls Secrets Manager itself. The security engineer wants to fix this with the least privilege possible. What should the engineer do?
- Attach an instance profile with secretsmanager:GetSecretValue to the Fargate tasks
- Change the role's trust policy principal from ecs-tasks.amazonaws.com to secretsmanager.amazonaws.com
- Grant secretsmanager:GetSecretValue on that secret to the task role
- Grant secretsmanager:GetSecretValue on that secret to the task execution role
Show the answer
Answer: D. Grant secretsmanager:GetSecretValue on that secret to the task execution role
The ECS/Fargate agent fetches secrets that the task definition injects, and it does so with the task execution role before the container starts, so that role needs the permission. The task role only gives credentials to the application code once it is running, and this code never calls Secrets Manager.
Question 6
A security engineer connects Okta to IAM Identity Center with SAML 2.0 and turns on SCIM provisioning. In the SCIM settings, the Username is mapped to each user's corporate email. Users and groups now appear correctly in Identity Center, but every user gets an error when they try to sign in to the AWS access portal. The Okta SAML app sends the employee ID as the NameID. What should the engineer do to fix sign-in?
- Edit the affected users in the Identity Center console so that their usernames match their employee IDs.
- Generate a new SCIM bearer token in Identity Center and paste it into the Okta AWS app.
- Change the Okta SAML app so that the NameID sends the same attribute that SCIM maps to Username (the corporate email).
- Run provision-permission-set with --target-type ALL_PROVISIONED_ACCOUNTS for every permission set.
Show the answer
Answer: C. Change the Okta SAML app so that the NameID sends the same attribute that SCIM maps to Username (the corporate email).
Sign-in fails when the SAML NameID doesn't match the attribute mapped to Username in SCIM, so the two have to be aligned in the IdP. Rotating the SCIM token won't help because sync is already working (users and groups appear). Editing users in the console isn't an option either, because once SCIM is on the IdP is the source of truth.
Question 7
A platform team lets developers create Lambda execution roles named app-* in account 111122223333. The developers' policy allows iam:CreateRole and iam:AttachRolePolicy only when iam:PermissionsBoundary equals the DevAppBoundary ARN, and it explicitly denies iam:DeleteRolePermissionsBoundary. The developers also have a broad allow on iam:*PolicyVersion* so they can manage their own customer managed policies. An audit warns that a developer could still give a new role administrator access. Which change closes this gap with the least impact on the developers' workflow?
- Attach DevAppBoundary as a permissions boundary to the IAM group that holds the developers
- Require developers to pass a session policy that excludes IAM actions whenever they assume their developer role
- Add an explicit Deny on iam:CreatePolicyVersion, iam:SetDefaultPolicyVersion, iam:DeletePolicyVersion and iam:DeletePolicy with the DevAppBoundary policy ARN as the resource
- Attach an SCP to the account that denies iam:CreateRole for every principal except the platform team
Show the answer
Answer: C. Add an explicit Deny on iam:CreatePolicyVersion, iam:SetDefaultPolicyVersion, iam:DeletePolicyVersion and iam:DeletePolicy with the DevAppBoundary policy ARN as the resource
The remaining escape route is to publish a new default version of DevAppBoundary that allows *, which raises the ceiling on every role that carries it, so denying version edits on that one policy ARN closes the gap and leaves the rest of the workflow alone. The SCP stops the escalation but also stops the self-service role creation the team wants, and you can't set a boundary on a group at all.
Question 8
A security engineer is adding an Amazon CloudFront distribution in front of an existing Application Load Balancer (ALB) in eu-west-2. The ALB's HTTPS listener already uses a public AWS Certificate Manager (ACM) certificate for www.example.com, issued in eu-west-2. When the engineer configures the distribution's custom SSL certificate, that certificate does not appear in the list. The goal is to serve viewers over HTTPS with a free, auto-renewing ACM certificate. What should the engineer do?
- Change the ALB listener's security policy to ELBSecurityPolicy-TLS13-1-2-2021-06 so the certificate becomes eligible for CloudFront
- Set the distribution's ViewerProtocolPolicy to redirect-to-https so CloudFront reuses the ALB's eu-west-2 certificate
- Request or import a certificate for www.example.com in ACM in us-east-1 and attach it to the distribution
- Request the certificate again in eu-west-2 using DNS validation instead of email validation
Show the answer
Answer: C. Request or import a certificate for www.example.com in ACM in us-east-1 and attach it to the distribution
CloudFront can only use ACM certificates from us-east-1. The ALB's certificate has to be in the ALB's own Region, so the eu-west-2 certificate stays on the ALB and a second certificate is needed in us-east-1. ViewerProtocolPolicy only controls how CloudFront handles HTTP requests from viewers. It does not let CloudFront use a certificate from another Region.
Question 9
A financial services company stores trade evidence in a versioned S3 bucket. Object Lock is set to a default retention of 1 year in governance mode. The legal team announces litigation of unknown length and needs a specific set of object versions kept until counsel releases them, even if their retention dates pass. Once the case ends, the versions must go back to normal clean-up. Which approach should the security engineer use?
- Place an Object Lock legal hold on the affected object versions, and limit s3:PutObjectLegalHold to the legal team's role
- Suspend the bucket's S3 Lifecycle expiration rule for the litigation prefix until counsel approves deletion
- Turn on S3 Replication to a bucket in another account so a copy of every version survives the case
- Change the affected versions to compliance mode with a 10-year retention period so they cannot be deleted before the case ends
Show the answer
Answer: A. Place an Object Lock legal hold on the affected object versions, and limit s3:PutObjectLegalHold to the legal team's role
A legal hold has no expiry date. It protects a version until someone with s3:PutObjectLegalHold removes it, which fits an audit or lawsuit of unknown length. Compliance mode with a guessed 10-year period can never be shortened, so the data would be kept long after the case ends and normal clean-up could not resume.
Question 10
A company runs a Control Tower landing zone on version 4.0. The security operations lead wants every drift event, such as a member account moved between OUs or a managed SCP detached, to open a ticket in the incident tool automatically. No one should have to check the console. What should the team set up?
- A CloudWatch metric filter and alarm on the organization CloudTrail log group in the log archive account, matching Organizations API calls
- An email or HTTPS subscription on the Control Tower SNS notification topic in the audit (aggregator) account
- An AWS Config aggregator rule in the aggregator account that forwards non-compliant evaluations to the incident tool
- An Amazon EventBridge rule in the management account that matches the detail-type "Drift Detected" and sends the events to the incident tool
Show the answer
Answer: D. An Amazon EventBridge rule in the management account that matches the detail-type "Drift Detected" and sends the events to the incident tool
In landing zone 4.0, Control Tower sends drift notifications to EventBridge in the management account with detail-type "Drift Detected", so you need a rule there to forward them. The SNS subscription is the closest distractor, but it describes how drift was delivered before 4.0, not how 4.0 does it.
The full timed mock
50 questions in 170 minutes, with a pass mark of 75%. Each paper is drawn from a bank of questions written for the mock alone, and these 10 are left out of it. Full access opens it, with every level of the course.
LevelUpWith is independent of the organisation that sets this exam. These questions were written for this course; they are not taken from the exam.