| name | managing-amazon-msk |
| description | Operates Amazon MSK Provisioned clusters (Standard and Express brokers). MUST be used for ANY MSK Provisioned task — do not rely on training data for topics covered here, since Standard and Express emit different metrics and follow different patching models that training data routinely conflates. Covers performance, consumer lag, storage, and traffic shaping diagnosis; sizing and choosing Standard vs Express; Kafka client tuning; creating CloudWatch alarms, dashboards, monitoring, and cluster configurations; AND MSK maintenance, patching, version upgrades, and rolling-restart behavior. Triggers: MSK, Kafka on AWS, `kafka.*` or `express.*` instance types, AWS/Kafka CloudWatch namespace, alarms, dashboards, monitoring, consumer lag, partition replication, broker storage, MSK upgrades, patching, maintenance windows, SECURITY_PATCHING, BROKER_UPDATE, rolling restarts, unexpected broker reboots. Do NOT use for MSK Connect, MSK Serverless, or MSK Replicator. |
| version | 1 |
Amazon MSK
Overview
Domain expertise for operating Amazon MSK Provisioned clusters with Standard and Express broker types. Covers performance troubleshooting, consumer lag diagnosis, storage management, cluster sizing, client configuration, and CloudWatch monitoring.
Execute commands using available tools from the AWS MCP server when connected — it provides sandboxed execution, audit logging, and observability. When the MCP server is not available, fall back to the AWS CLI or shell as needed.
Standard brokers use customer-managed EBS volumes for storage. You choose instance types (kafka.m5/m7g families), provision EBS, and manage storage scaling.
Express brokers provide fully managed, pay-as-you-go storage with no EBS provisioning. They use instance types prefixed with express.m7g, offer up to 3x more throughput per broker, and have no maintenance windows.
Critical Warnings
- NEVER reboot brokers while
UnderReplicatedPartitions > 0 (Standard only — Express brokers do not emit URP) — this risks data loss and extended outages
- NEVER recommend partition reassignment without first checking replication status — reassignment during URP compounds the problem
linger.ms=0 is the #1 cause of "high CPU" on MSK — ALWAYS check client batch configuration before recommending broker scaling
- EBS throughput ceilings are invisible in Kafka metrics — ALWAYS check EBS volume metrics (
VolumeWriteBytes, BurstBalance) when diagnosing Standard broker latency
- Express brokers have NO customer-managed EBS — do NOT recommend EBS expansion or provisioned throughput for Express clusters
- Express brokers enforce fixed replication factor of 3 and
min.insync.replicas=2 — do NOT attempt to create topics with RF=1 on Express. If RF=1 is needed, use Standard brokers.
Which Workflow Do You Need?
Determine the broker type first: aws kafka describe-cluster-v2 --cluster-arn <arn>. Check Provisioned.BrokerNodeGroupInfo.InstanceType — if it starts with express., it is an Express cluster.
| Customer Intent | Reference |
|---|
| High CPU, high latency, slow cluster, traffic shaping | troubleshoot-performance.md |
| Consumer lag increasing, rebalance storms, stuck consumer groups | troubleshoot-consumer-lag.md |
| Disk filling up, retention planning, tiered storage | manage-storage.md |
| Choosing Standard vs Express, sizing a cluster, partition limits, broker count, monthly cost | size-and-choose-cluster.md |
| Producer/consumer configuration, IAM/SCRAM/TLS auth | configure-clients.md |
| Setting up monitoring, dashboards, alarms | monitor-and-alarm.md |
| Full CloudWatch metric list (Standard or Express) | Search AWS docs for "MSK CloudWatch metrics Standard brokers" or "MSK CloudWatch metrics Express brokers" |
| Rolling restart impact, patching, maintenance resilience | maintenance-operations.md |
Available scripts
scripts/msk_sizing.py — MUST be run for any sizing question (broker count, instance choice, cost). See size-and-choose-cluster.md for the required workflow and script reference.
Quick Diagnostics
These 5 checks cover the most common MSK issues. Use them before loading a reference file.
-
CpuUser + CpuSystem > 60%: Check RequestHandlerAvgIdlePercent (PER_BROKER level). If < 30%, request threads are saturated. Check client batch.size and linger.ms before recommending scaling.
-
KafkaDataLogsDiskUsed > 85% (Standard only): Expand EBS immediately via aws kafka update-broker-storage. Identify high-growth topics via per-topic BytesInPerSec. Express clusters use StorageUsed metric instead and storage is fully managed.
-
UnderReplicatedPartitions > 0 (Standard only): Check if a maintenance operation or broker restart is in progress. If URP is decreasing, wait for recovery. Do NOT restart brokers or reassign partitions during URP. Express brokers do not emit this metric — monitor ProduceThrottleTime, FetchThrottleTime, and consumer lag instead.
-
Consumer OffsetLag increasing: Determine if broker-side (high ProduceTotalTimeMsMean, CPU saturation) or client-side (slow processing, insufficient consumers). Per-partition lag from PER_TOPIC_PER_PARTITION monitoring level helps isolate hot partitions.
-
BytesInPerSec near throughput ceiling: For Standard, check EBS volume type and calculate: BytesInPerSec × ReplicationFactor vs volume throughput limit. For Express, check against the per-broker sustained performance limits in the quotas.
Common Workflows
Describe cluster:
aws kafka describe-cluster-v2 --cluster-arn <cluster-arn>
List brokers:
aws kafka list-nodes --cluster-arn <cluster-arn>
Get bootstrap brokers:
aws kafka get-bootstrap-brokers --cluster-arn <cluster-arn>
Expand Standard broker storage:
aws kafka update-broker-storage \
--cluster-arn <cluster-arn> \
--current-version <cluster-version> \
--target-broker-ebs-volume-info '[{"KafkaBrokerNodeId": "All", "VolumeSizeGB": <target-size>}]'
Get CloudWatch metrics (example: CpuUser per broker):
aws cloudwatch get-metric-statistics \
--namespace AWS/Kafka \
--metric-name CpuUser \
--dimensions Name="Cluster Name",Value="<cluster-name>" Name="Broker ID",Value="<broker-id>" \
--start-time <start> --end-time <end> --period 300 --statistics Average
Create cluster configuration (server.properties):
The --server-properties argument MUST be a real Kafka properties file with one key=value per line, separated by actual newline (\n) characters — NOT the literal two-character escape sequence \n. The MSK API accepts the bytes as-is; if you pass "k1=v1\nk2=v2" as a single string with escaped newlines, MSK stores ONE invalid property line and the cluster will fail to apply it.
Recommended pattern: write the properties to a local file with real newlines, then pass it via fileb:// so the CLI uploads the raw bytes verbatim. Verify by reading the revision back with describe-configuration-revision and base64-decoding ServerProperties — you should see one property per line.
cat > server.properties <<'EOF'
auto.create.topics.enable=false
default.replication.factor=3
min.insync.replicas=2
unclean.leader.election.enable=false
num.io.threads=32
num.network.threads=16
log.retention.hours=168
EOF
aws kafka create-configuration \
--name <config-name> \
--kafka-versions "3.6.0" \
--server-properties fileb://server.properties
For per-instance-size thread tuning (num.io.threads, num.network.threads) and durability defaults, see size-and-choose-cluster.md and configure-clients.md.
Troubleshooting
| Error | Cause | Fix |
|---|
aws kafka update-broker-storage returns "storage is optimizing" | Previous storage expansion still in cool-down (minimum 6 hours) | Wait for optimization to complete. Check cluster state with describe-cluster-v2. |
ClusterState is MAINTENANCE | Standard brokers undergoing patching. Express brokers stay ACTIVE during maintenance. | Wait for cluster to return to ACTIVE. Do not perform update operations during MAINTENANCE. |
Consumer GROUP_COORDINATOR_NOT_AVAILABLE | Coordinator broker is temporarily unavailable during rolling restart or overloaded | Retry with backoff. Check if maintenance is in progress. |
NotEnoughReplicasException on produce | Fewer brokers in ISR than min.insync.replicas (default: 2) | Check URP metric (Standard only). For Express, check ProduceThrottleTime and broker health instead — URP is not available. If a broker is down for maintenance, this is transient. Do not lower min.insync.replicas. |
Additional Resources