| CLUSTER INFORMATION |
| Cluster State | The current operational state of the MSK cluster.
Possible Values: ACTIVE, CREATING, UPDATING, DELETING, FAILED, MAINTENANCE, REBOOTING_BROKER, HEALING |
| Cluster Type | The type of backend cluster. |
| Broker Type | The type of broker used for the cluster.
Possible Values: STANDARD, EXPRESS |
| Metadata Mode | The configuration of how cluster metadata is managed.
Possible Values: KRaft, ZooKeeper
Note: Applicable only for STANDARD broker type. Metadata mode configuration is supported only for STANDARD clusters. EXPRESS clusters use AWS-managed serverless configuration by default and do not provide an option to select or modify the metadata mode. |
| Number of Brokers | The total number of broker nodes currently provisioned in the cluster. |
| Zookeeper Session State | The connection status of the ZooKeeper session for brokers at the time of polling. This represents the worst state observed among all brokers.
Possible Values:NOT_CONNECTED, ASSOCIATING, CONNECTING, CONNECTED_READ_ONLY, CONNECTED, CLOSED, AUTH_FAILEDNote: Applicable only for STANDARD broker type with ZooKeeper as the metadata mode. ZooKeeper session state is relevant only when the cluster uses ZooKeeper for metadata management and is not applicable for KRaft or EXPRESS clusters. |
| ACTIVE CONTROLLERS |
| Active Controllers | The minimum number of active controllers managing the cluster between the polling interval. This helps track if the cluster ever loses its active controller, which is critical to prevent control plane failures. |
| KAFKA DATALOGS DISK UTILIZATION |
| Kafka Datalogs Disk Utilization | The average disk space utilized for Kafka data logs at the time of polling (in %).
Note: Applicable only for STANDARD broker type. Disk utilization metrics are applicable only to STANDARD clusters. EXPRESS clusters use serverless elastic storage and therefore do not report disk utilization. |
| OFFLINE PARTITIONS |
| Offline Partitions | The maximum number of partitions in the cluster that are offline between the polling interval. This helps track inactive or down partitions, indicating data unavailability or broker outages.
Note: Applicable only for STANDARD broker type. Partition leadership and offline state metrics are available only for STANDARD clusters. EXPRESS clusters are fully AWS-managed and do not expose partition leadership or offline state details to users. |
| GLOBAL PARTITIONS |
| Global Partitions | The maximum number of partitions in a cluster, excluding replicas, between the polling interval. This helps track the highest total partition count at any time, essential for resource allocation and capacity planning. |
| GLOBAL TOPICS |
| Global Topics | The maximum number of topics across all brokers in a cluster between the polling interval. This helps track the peak topic count, which is important for assessing capacity requirements and workload spikes. |
| CLIENT CONNECTIONS |
| Client Connections | The average number of clients that are actively connected to the brokers in the cluster during the polling interval. |
| ZOOKEEPER REQUEST MEAN LATENCY |
| ZooKeeper Request Mean Latency | The mean latency for Apache ZooKeeper requests during the polling interval (in ms).
Note: Applicable only for STANDARD broker type with ZooKeeper as the metadata mode. This metric is not applicable in KRaft mode or for EXPRESS clusters. |
| EXPRESS CLUSTER STORAGE USED |
| Express Cluster Storage Used | The average amount of storage used across all partitions in the cluster, excluding replicas between the polling interval.
Note: Applicable only for EXPRESS broker type. Express Cluster Storage Used is specific to EXPRESS clusters, which use AWS-managed serverless elastic storage, and is not applicable to STANDARD clusters with EBS-backed storage. |
