The Poison-Pill MessageKafka · medium · ~10 minThe payments-enricher group processed normally until 14:02, then partition 7's lag froze in place and started growing while the other 11 partitions stayed at zero lag. The consumer assigned to partition 7 is crash-looping.Every restart crashes at the same place. Find the message that's killing it — and design the permanent fix.Evidence

Metrics

Consumer Logs

Config

Checkpoints

Why does restarting never help?

The permanent consumer-side fix?

The malformed messages came from a partner's bad deploy. What belongs on the producer/platform side, and how do you narrate the whole fix in an interview?