Question
Starting 03:50, the `notifications` Kafka consumer group (24 partitions, 12 instances) enters a state where lag oscillates wildly — climbing to 2M, dropping to 200k, climbing again — and never stabilizes. Dashboards: the group's `rebalance-rate-per-hour` JMX metric is pinned high, instances log `Member X sending LeaveGroup ... consumer poll timeout has expired` and `Attempt to heartbeat failed ... group is rebalancing` in a loop. CPU is moderate, no OOMs. Recent context: yesterday a new feature added a per-message call to a templating service whose p99 is ~600ms, and message volume is normal. `max.poll.records=500`, `max.poll.interval.ms=300000` (default 5 min), `session.timeout.ms=10000`. How do you triage and mitigate?
Stop the bleeding first (mitigate), then form hypotheses from real signals. Separate root cause from symptom, communicate status as you go, and close with what prevents a repeat.