Community Q&A
SCS-C03 — frequently asked questions
Community-vetted answers to 35 common questions about this exam, each traced back to a specific practice question.
Questions from real practice questions
Each Q&A comes from a specific question — follow the link for the full analysisHow Can the Security Team Trace a Former Employee's Access Key Activity in AWS?
Why isn't Amazon Athena the better choice for investigating this access key?
Athena is right only after events age past 90 days into an S3 log archive, and it requires a configured trail plus a table definition — setup the scenario never mentions.
Can AWS Config show which actions a specific user or access key performed?
No. AWS Config tracks resource configuration changes and compliance, not a searchable per-principal history of API calls; CloudTrail Event History is the tool for that.
How to Authenticate External Users Across ECS Microservices?
Why isn't the per-microservice IAM role part of the ordered Cognito/API Gateway steps?
IAM task roles give ECS microservices temporary AWS permissions, but they do not authenticate external web users. The question's ordered flow is specifically the user-login and token-validation chain.
Does the API Gateway Lambda authorizer replace Amazon Cognito user pools?
No; Cognito user pools authenticate users and issue tokens, while the Lambda authorizer validates those tokens at API Gateway before requests reach the ECS microservices.
How to enforce MFA for AWS CLI EC2 commands?
Why do long-term AWS CLI access keys fail when the IAM policy requires MFA?
Long-term access keys are not MFA-authenticated, so aws:MultiFactorAuthPresent stays false and the deny applies. STS GetSessionToken returns temporary MFA-backed credentials.
Can adding sts:AssumeRole to NotAction in the MFA deny policy solve CLI access?
It can shift enforcement to a role trust policy, but it changes the architecture and can weaken the original deny. Option B's GetSessionToken flow directly supplies MFA-authenticated CLI credentials.
How to Provide Keyless SSH Access to EC2 and On-Premises Servers?
Why is a bastion host wrong when Session Manager is available?
A bastion still needs SSH keys, an always-on inbound SSH path, and separate access for on-premises servers, so it conflicts with the no-key and isolation requirements.
Does Session Manager work for on-premises servers in this scenario?
Yes. The SSM Agent can be activated on on-premises servers as a managed instance, allowing IAM-controlled sessions without inbound ports or SSH keys.
How to Detect EKS Audit Log and Runtime Risks with Email Alerts?
Why does GuardDuty EKS Protection need Runtime Monitoring as well?
EKS Protection analyzes Kubernetes audit logs for control-plane risks, while Runtime Monitoring observes container and node activity such as file access, process execution, and network connections.
Why can't Security Hub EKS controls replace GuardDuty here?
Security Hub standards check configuration posture against EKS controls; they do not analyze EKS audit logs or generate runtime findings for OS, network, and file events.
How to Create CloudHSM-Backed KMS Keys and Audit Their Use?
Why is Amazon GuardDuty wrong for auditing KMS key use here?
GuardDuty detects threats and anomalous behavior; it does not provide the authoritative record of KMS API and key-use events that CloudTrail does.
Why not use Amazon S3 with a CloudHSM custom key store to create keys?
Amazon S3 is object storage and cannot create or control KMS keys; only AWS KMS integrates with a CloudHSM-backed custom key store for that purpose.
How Should Auto Scaling Applications Preserve Logs for One Year?
Why doesn't EFS with a daily log copy meet the durability requirement?
The application still writes to local EBS first, so a scale-in before the daily copy loses up to a day of logs. EFS is durable, but the copy process is not continuous.
Can CloudWatch Logs retention be set to one year?
Yes. CloudWatch Logs retention is configured per log group, and you can set it to 365 days or another period that satisfies the audit requirement.
How to recover an RDS cluster to the last known good version after an attack?
Why is restoring the snapshot closest to the attack time not precise enough?
Snapshots are discrete recovery points and may be hours away from the attack time. Point-in-time recovery can restore to a specific second just before the attack, minimizing data loss.
Why not set point-in-time recovery to 14 days ago?
That restores to the start of the retention window, not the last known good state before the attack. It would discard valid changes made after that earlier point and may hit the retention boundary.
How Do You Disable a Compromised IAM Identity Center User and Audit Activity?
Why is disabling a management-account IAM user wrong for Identity Center access?
IAM users are per-account and separate from IAM Identity Center users; disabling the management-account IAM user does not stop the compromised Identity Center user from signing in to other assigned accounts.
Can CloudWatch Logs Insights query CloudTrail logs stored in the S3 bucket?
No. Logs Insights queries CloudWatch Logs log groups, not S3 objects; the organizational event data store is queried through CloudTrail Lake instead.
How to Find S3 Buckets with Sensitive Data Using Macie with Least Overhead?
Why doesn't a Lambda triggered by S3 upload events satisfy the Macie requirement?
It only scans objects uploaded after the trigger is configured, ignores existing objects, and requires per-bucket event wiring and code maintenance — far more overhead than Macie automated discovery.
Does Macie automated discovery perform a full scan of every object?
No. It uses intelligent sampling to evaluate buckets and objects on a schedule, which is why you then run targeted full classification jobs only on the buckets it flags as sensitive.
Which Network ACL Rule Set Allows TLS 443 and Denies MySQL 3306?
Why does option B allow inbound TCP 1024–65535 instead of TCP 443 for the TLS response?
NACLs are stateless: the return packet for an outbound TLS connection is destined to the EC2 instance's ephemeral source port, not to port 443.
Why is option C invalid if it also includes a deny for MySQL port 3306?
Its inbound rule 100 allows TCP 1024–65535 first, and NACLs evaluate the lowest rule number first, so MySQL 3306 is allowed before rule 200 can deny it.
How Should Secrets Manager Secrets Be Encrypted and Replicated to us-west-1?
Why can't option D use the us-east-1 customer managed KMS key for the us-west-1 replica?
A standard KMS customer managed key is Regional and cannot encrypt data in us-west-1. D only works if the key is explicitly multi-Region with a replica key in us-west-1.
Does Secrets Manager replication require a customer managed KMS key?
No. Secrets Manager can replicate secrets encrypted with the AWS managed aws/secretsmanager key, and the destination replica can use the destination Region's AWS managed key.
How Should You Handle Leaked IAM Access Keys?
Why is IAM Access Analyzer wrong for checking who used the exposed key?
IAM Access Analyzer evaluates resource policies to find external access; it does not log API calls or identify the principal behind a credential's use. CloudTrail or the credential report is appropriate for usage review.
Does deactivating the access key delete it permanently?
No. Deactivation sets the key to Inactive so API calls fail, but the key remains listed; you can delete it later after investigation.
How should a secured S3 CloudFormation template be shared across AWS Organizations?
Why can't AWS RAM share the CloudFormation template across the organization?
AWS RAM shares supported AWS resources, not CloudFormation templates. To distribute an approved template for repeated deployments, use a Service Catalog product in a shared portfolio.
How does Service Catalog keep the secured S3 configuration standardized?
End users launch the product from the portfolio but cannot modify the underlying CloudFormation template. Each launch creates a new S3 bucket with the approved secure settings.
How to Recover an EBS Volume After KMS Key Material Deletion?
Why can't new KMS key material decrypt the existing EBS volume?
The EBS data key was wrapped by the original customer managed key and its material; different key material creates a different cryptographic key that cannot unwrap the existing ciphertext.
Does restoring a pre-deletion EBS snapshot avoid the unusable KMS key?
No. An encrypted snapshot still depends on the same CMK; without the original material it cannot be restored, and it may omit data written after the snapshot.
How to enforce a third-party WAF managed rule group across all accounts and Regions?
Why can't an SCP add a WAF managed rule group to existing web ACLs?
SCPs only constrain IAM permissions; they cannot modify or populate AWS WAF web ACL rule groups, so option B cannot enforce the third-party managed rule group.
Does Security Hub CSPM remediate a missing WAF rule group?
No. Security Hub CSPM can generate findings for missing controls but does not automatically insert the third-party managed rule group into every web ACL.
Lambda S3 Read Failure: Fix the Bucket Policy Resource ARN
Why doesn't a bucket policy granting s3:GetObject on the bucket ARN work?
s3:GetObject is an object-level action and requires an object ARN such as arn:aws:s3:::DOC-EXAMPLE-BUCKET/*. The bucket ARN alone only authorizes bucket-level actions like s3:ListBucket.
Should the Principal be lambda.amazonaws.com so MyLambdaFunction can read S3?
No. Lambda signs S3 requests with its execution role, so grant access to that role ARN or an account principal, and ensure the role's identity policy also allows s3:GetObject.
Would removing the bucket policy Condition always fix Lambda access?
No. A condition may still be valid, and removing it can broaden access. First verify the condition key is present for the actual Lambda-to-S3 request and fix the resource or principal mismatch.
← Back to SCS-C03 AWS Certified Security Specialty Study Guide