DBS-C01 Monitoring and Troubleshooting Practice Test 20 — 15 Questions

Question 1

A company is running a MongoDB-compatible Amazon DocumentDB cluster with one writer and two readers. The application writes a large amount of data during batch processing, and after a batch completes, the writer's CPU is high, and the readers have significant replica lag. The team wants to reduce replica lag without affecting the batch performance. What should they do?

Choose an answer to begin — your selection is scored in the full session.

Accepted Answer

Increase the instance size of the readers to improve apply throughput. Replica lag in Amazon DocumentDB occurs when readers are unable to apply writes from the writer's oplog quickly enough. Increasing the instance size of the readers provides more CPU and memory resources, improving their apply throughput and reducing lag without affecting the writer's batch performance. Option A is incorrect because changing the storage type to Provisioned IOPS may improve I/O but does not directly address the reader's apply capacity. Option C is incorrect because reducing batch size would lower the write rate and degrade batch performance, which the team wants to avoid. Option D is incorrect because increasing the writer's instance size may help the writer handle more writes, but it does not help readers apply those writes faster; it could even increase the write rate and worsen replica lag.

Answer

Change the storage type to Provisioned IOPS on all instances

Answer

Reduce the batch size to lower the write rate

Answer

Increase the instance size of the writer to handle the batch faster

Question 2

A developer is receiving timeout errors when connecting to an Amazon ElastiCache for Redis cluster from an Amazon EC2 instance. The security group for the EC2 instance allows outbound traffic to the Redis cluster's security group on port 6379. The Redis cluster's security group does not allow inbound traffic from the EC2 instance. What is the most likely cause of the timeout?

Accepted Answer

The Redis cluster security group does not have an inbound rule allowing traffic from the EC2 security group on port 6379. Security groups are stateful, meaning that if an outbound rule allows traffic, the return traffic is automatically allowed. However, inbound traffic must be explicitly permitted by the target security group. In this scenario, the EC2 instance's security group allows outbound traffic to the Redis cluster's security group on port 6379, but the Redis cluster's security group does not have an inbound rule allowing traffic from the EC2 security group on that port. Therefore, the connection is blocked, causing timeout errors. Option A is incorrect because network ACLs are stateless and would affect both inbound and outbound traffic, but the issue is specifically with security groups. Option C is incorrect because subnet route tables control network routing, not port-level access. Option D is incorrect because ElastiCache for Redis clusters are accessible within the same VPC if proper security group rules are configured.

Answer

The network ACL for the Redis subnet is blocking inbound traffic on port 6379

Answer

The subnet route table does not have a route to the Redis cluster

Answer

The Redis cluster is not accessible from within the same VPC

Question 3

A company uses Amazon RDS for PostgreSQL with Multi-AZ deployment. The primary instance fails, and a failover occurs. After the failover, the application is still unable to connect to the database endpoint. The database administrator checks the RDS console and sees that the new primary is in 'available' state. What should the administrator do next to diagnose the connectivity issue?

Accepted Answer

Check the DNS resolution of the RDS endpoint from the application server. In an RDS Multi-AZ failover, the CNAME record for the DB endpoint is updated to point to the new primary's IP address, but DNS caching on the application server (or its JVM/.NET resolver) can retain the old IP for the TTL duration. Since the console shows the new primary as 'available', the failover itself succeeded, so the next diagnostic step is to verify that the application is resolving the endpoint to the new primary's IP. This is the classic post-failover connectivity issue documented by AWS.

Answer

Verify that the security group for the RDS instance allows inbound traffic from the application

Answer

Check if the subnet group for the RDS instance is correctly configured

Answer

Restart the RDS instance to force a new connection

Question 4

A database administrator is troubleshooting a performance issue on an Amazon Aurora MySQL cluster. The application is experiencing high latency on write operations. Which TWO CloudWatch metrics should the administrator analyze to identify the root cause?

Accepted Answer

DMLThroughput. CommitLatency (option C) is correct because it measures the time spent committing write transactions in Aurora MySQL, directly reflecting the latency the application experiences on write operations and helping pinpoint whether commit processing is the bottleneck. DMLThroughput (option B) is correct because it reports the number of INSERT, UPDATE, and DELETE statements executed per second, which is essential to correlate write workload volume with the observed latency and determine if throughput spikes are causing the slowdown. ReadLatency (option A) and SelectLatency (option D) are incorrect because they measure read-side performance, not write operations, so they would not explain high write latency. FreeableMemory (option E) is incorrect because it indicates available memory rather than write latency, and while memory pressure can affect performance, it is not a direct write-latency metric for diagnosing this issue.

Answer

ReadLatency

Answer

SelectLatency

Answer

FreeableMemory

Question 5

A company is using Amazon DynamoDB for a gaming leaderboard. The table has a partition key of 'game_id' and a sort key of 'score'. The table is configured with on-demand capacity. During a major tournament, the application experiences high latency and some requests return 'ProvisionedThroughputExceededException' errors. The CloudWatch metric 'ThrottledRequests' spikes. The application uses a single partition key for all writes during the tournament (game_id = 'tournament_final'). What is the most likely cause of the throttling, and what is the best solution?

Accepted Answer

The application is using a single partition key, causing all writes to go to one partition. The team should redesign the partition key to distribute writes across multiple partitions. The throttling is caused by using a single partition key ('tournament_final') for all writes, which concentrates all write activity on one partition. DynamoDB partitions have a maximum throughput limit (e.g., 1,000 WCU per partition for on-demand), so a single partition key cannot scale beyond that limit. The best solution is to redesign the partition key to distribute writes across multiple partitions, such as using a composite key or adding a random suffix.

Answer

The application is using a single partition key, causing all writes to go to one partition. The team should implement DAX to cache writes

Answer

The table has a global secondary index that is throttling writes; the team should remove the GSI

Answer

The table is using on-demand capacity, which has a maximum throughput limit per partition; the team should switch to provisioned capacity with auto scaling

Question 6

A company runs an Amazon Redshift cluster with 8 dc2.large nodes for its data warehouse. The data engineering team loads data daily using COPY commands from S3. Recently, the load times have increased significantly. The cluster's CloudWatch metric 'CPUUtilization' is high during the load. The administrator runs the STL_LOAD_ERRORS table and finds no errors. The SVL_S3LOG shows that the COPY command is scanning many small files. The data in S3 is stored as 10,000 small CSV files (each ~100 KB). Which action will MOST improve the COPY performance?

Accepted Answer

Consolidate the small files into fewer, larger files (e.g., 100 files of 10 MB each). Amazon Redshift COPY performance is heavily influenced by the number and size of files in S3. Loading from many small files (10,000 files of ~100 KB) causes excessive overhead because each file is processed individually, leading to high CPU utilization and slow loads. Consolidating into fewer, larger files (e.g., 100 files of 10 MB each) allows Redshift to parallelize the load more efficiently across slices, dramatically improving performance.

Answer

Use the MANIFEST option to specify the files explicitly

Answer

Use the JSON format instead of CSV to reduce parsing overhead

Answer

Change the table's distribution style to ALL to avoid data redistribution

Question 7

A developer is troubleshooting an application that writes to an Amazon ElastiCache for Redis cluster. The application occasionally fails with 'OOM command not allowed when used memory > maxmemory'. What is the most likely cause?

Accepted Answer

The cluster's maxmemory-policy is set to noeviction.. The error 'OOM command not allowed when used memory > maxmemory' occurs when Redis has reached its configured memory limit and the eviction policy is set to 'noeviction', which prevents any further writes. Option B is correct because 'noeviction' causes Redis to return an error instead of evicting keys. Option A is incorrect because the 'maxclients' limit produces a different error ('max number of clients reached'). Option C is incorrect because keyspace notifications do not affect memory limits. Option D is incorrect because cross-slot commands relate to cluster mode, not memory errors.

Answer

The cluster's maxclients limit has been reached.

Answer

The cluster has too many keyspace notifications enabled.

Answer

The cluster is in cluster mode and cross-slot commands are used.

Question 8

A database specialist is troubleshooting a performance issue on a self-managed PostgreSQL database that they plan to migrate to Amazon RDS. The database has a high number of 'idle in transaction' connections. What is the impact of these connections on the database?

Accepted Answer

They hold locks and prevent cleanup of dead tuples, leading to bloat.. Idle-in-transaction connections hold locks and prevent PostgreSQL's autovacuum from cleaning up dead tuples, leading to table bloat and performance degradation. Option A is wrong because idle transactions do not cause constant polling; CPU usage remains low. Option C is wrong because they do not prevent new connections from being established unless the max_connections limit is reached. Option D is wrong because idle transactions do not significantly increase write-ahead logging.

Answer

They increase CPU usage due to constant polling.

Answer

They prevent new connections from being established.

Answer

They cause increased disk I/O from write-ahead logging.

Question 9

A company uses Amazon ElastiCache for Redis. They want to monitor cache hit ratio. Which TWO metrics should be used to calculate the cache hit ratio?

Accepted Answer

CacheHits. Cache hit ratio is calculated as CacheHits divided by (CacheHits + CacheMisses), so option B (CacheHits) and option D (CacheMisses) are the two metrics needed. CacheHits counts the number of successful read-only key lookups in the main dictionary, while CacheMisses counts the number of unsuccessful read-only key lookups, and together they represent the total number of read attempts. Option A (GetTypeCmds) counts the total number of read-only type commands and does not distinguish hits from misses, so it cannot be used to compute the ratio. Option C (Evictions) tracks keys evicted due to maxmemory limits, which relates to memory pressure rather than hit/miss performance. Option E (CurItems) reports the current number of items stored in the cache, which is a capacity metric, not a hit/miss metric.

Answer

GetTypeCmds

Answer

Evictions

Answer

CurItems

Question 10

A retail company uses Amazon RDS for PostgreSQL as the backend for its e-commerce platform. During a flash sale, the database experienced high CPU utilization and increased the number of active connections. The application team reported that some queries timed out. The database specialist reviewed the slow query log and found that several queries were performing sequential scans on large tables due to missing indexes. The specialist created the necessary indexes, but the issue persists for some queries. Upon further investigation, the specialist notices that the query planner is still choosing sequential scans for some queries. What should the database specialist do to ensure the query planner uses the indexes?

Accepted Answer

Run the ANALYZE command to update table statistics.. The query planner relies on table statistics to estimate the cost of different execution plans. If statistics are stale, the planner may incorrectly estimate the number of rows returned by a sequential scan versus an index scan, leading it to choose a sequential scan even when an index exists. Running ANALYZE updates these statistics, allowing the planner to make better decisions.

Answer

Increase maintenance_work_mem to speed up index creation.

Answer

Decrease the random_page_cost to make indexes more attractive.

Answer

Set enable_seqscan to off to force index usage.

Question 11

A startup is using Amazon ElastiCache for Redis to cache session data. They deployed a single Redis node (cache.t3.micro) in us-west-2. The application reports high latency when reading session data. CloudWatch metrics show CPUUtilization at 90% and Evictions at 100 per minute. The cache hit ratio is 80%. The database specialist suspects the node is overloaded. What should the specialist do to improve performance?

Accepted Answer

Scale up to a larger node type, such as cache.m5.large.. The CloudWatch metrics show CPUUtilization at 90% and Evictions at 100 per minute on a single cache.t3.micro node, indicating the node is both CPU- and memory-constrained. Scaling up to a larger node type such as cache.m5.large provides more CPU and memory, directly addressing the overload and reducing evictions. This is the correct first action for a single-node Redis cache under resource pressure.

Answer

Add a read replica to offload read traffic.

Answer

Enable cluster mode and add more shards.

Answer

Decrease the TTL for session keys to reduce memory usage.

Question 12

A financial services company is using Amazon Aurora MySQL as its primary database. The database has a table 'transactions' that receives high inserts during business hours. The table is partitioned by date. Recently, the application team noticed an increase in lock wait timeouts. The database specialist reviewed the InnoDB status and found that there are frequent gap locks on the 'transaction_date' column. The isolation level is REPEATABLE READ. What should the specialist do to reduce lock waits while maintaining data consistency?

Accepted Answer

Change the transaction isolation level to READ COMMITTED.. Under REPEATABLE READ, InnoDB uses next-key locks (record + gap locks) to prevent phantom reads, which causes gap locks on the transaction_date column during inserts. Switching to READ COMMITTED disables gap locking (except for foreign-key and duplicate-key checks), reducing lock waits while still maintaining statement-level consistency. This is the standard Aurora MySQL recommendation for high-concurrency insert workloads.

Answer

Add a secondary index on transaction_date.

Answer

Increase the innodb_lock_wait_timeout parameter.

Answer

Modify the partitioning key to use a hash-based partition.

Question 13

A database administrator notices that an Amazon RDS for MySQL DB instance is experiencing high CPU utilization and increased latency during peak hours. The administrator wants to identify the queries causing the issue. Which TWO actions should be taken to diagnose the problem? (Select TWO.)

Accepted Answer

Enable Performance Insights on the DB instance.. Enabling Performance Insights provides database performance metrics including wait events and SQL query details, allowing identification of queries causing high CPU and latency. Option C: Enabling the slow query log captures queries that exceed a specified duration threshold, helping to pinpoint slow-running queries. Option B (Amazon DevOps Guru for RDS) is an automated service for anomaly detection and performance insights, not a direct method to identify specific queries. Option D (RDS Proxy) manages connection pooling and improves scalability, but does not identify problematic queries. Option E (RDS Enhanced Monitoring) provides OS-level metrics, not query-level details.

Answer

Enable Amazon DevOps Guru for RDS.

Answer

Use RDS Proxy to manage connections.

Answer

Enable RDS Enhanced Monitoring.

Question 14

A company is running Amazon RDS for MySQL and notices that the database CPU utilization is consistently above 80% during peak hours. The application performance is degrading. Which action should be taken first to troubleshoot the issue?

Accepted Answer

Enable Performance Insights to identify the queries causing high CPU usage.. Enabling Performance Insights is the correct first step because it provides detailed visibility into database load, helping identify the specific SQL queries or wait events causing high CPU utilization. This diagnostic information is essential before taking remedial actions like scaling or adding replicas.

Answer

Increase the instance size of the RDS instance immediately.

Answer

Create a read replica to offload read traffic.

Answer

Switch the database engine to Amazon Aurora for better performance.

Question 15

A company has an Amazon DynamoDB table with on-demand capacity. Users report that write requests are occasionally throttled during peak hours. The application uses the AWS SDK and retries with exponential backoff. Which monitoring approach should be used to identify the cause of throttling?

Accepted Answer

Monitor CloudWatch metrics for ThrottledRequests and ConsumedWriteCapacityUnits.. CloudWatch metrics such as ThrottledRequests and ConsumedWriteCapacityUnits directly indicate throttling events and capacity usage, enabling identification of throttling causes even with on-demand capacity. Option A is incorrect because AWS CloudTrail logs API calls but does not provide detailed throttling metrics at the item level. Option B is incorrect because VPC Flow Logs capture network traffic, not database-level throttling. Option D is incorrect because DynamoDB Streams capture item-level changes, not throttling events.

Answer

Use AWS CloudTrail to monitor PutItem API calls for throttling errors.

Answer

Analyze VPC Flow Logs to check for network congestion.

Answer

Enable DynamoDB Streams and process records to monitor throttled requests.