Monitor consumer group lag with operational diagnosis.

Checks committed offsets vs log-end-offsets for each partition and generates diagnosis based on actual incident patterns:

• Flink groups with "no active members" is NORMAL (checkpoint-based offset management)

• Flink groups with no active members AND growing lag = likely Flink crash (matches 2026-02-21 incident: MySQL DELETE → tombstone → NPE → 50h outage)

• Non-Flink groups with no active members = consuming application is down

• Uneven lag distribution = possible hot partition or stuck consumer

Args: group: Consumer group ID. Empty = all groups.

Get topic details: partitions, replicas, ISR, configs, and health diagnosis.

Checks replication health:

• ISR < min.insync.replicas = writes FAILING (critical)

• ISR < replication.factor but >= min.insync.replicas = degraded (warning)

• Retention policy info (size-based deletion warnings)

Args: topic: Topic name to inspect

Check Kafka cluster health: brokers, controller, under-replicated partitions.

Detects:

• Missing brokers (expected 3 for this cluster)

• ClusterIdMismatch pattern (Docker volume recreation issue)

• Under-replicated partitions across all topics

List all Kafka topics with partition count, replication factor, and descriptions.

Built-in descriptions for known topics (CDC pipeline, Circuit Connect).

Args: include_internal: Include internal topics starting with '_' (default: False)