Kafka
Know why the consumer stopped, not just that it is behind.
The consumer, the topic, the schema change and the team behind it, with the evidence under every claim
The finding
See the stall your broker dashboard calls healthy.
A frozen offset, the schema change that caused it and the minutes in between, laid out as one timeline
Evidence
You do not have to re-run the queries yourself.
Offsets, consumer group state and the log silence in between, each one showing the query it came from
Memory
The same schema change does not stall you twice.
A narrative and the learnings drafted when the investigation closes, each one approved by a person before it lands
Frequently asked questions.
How does an investigation start?
From an alert, from Slack, or by asking. The investigation names the consumer, the topic, and the change.
Does this replace kafka-exporter or Burrow?
No. Those hand you a lag number. This names the service, the team, and the change behind it.
Does NOFire read the messages?
No. It reads cluster metadata: topics, groups, offsets, schema versions. It never consumes your records.
Does it change production?
No. It does not revert, restart, or skip a poison pill. You decide what happens next.
What Kafka does this cover?
MSK, Confluent, Strimzi, or self-hosted. Same investigations.
How is this different from broker dashboards?
Brokers can stay green while a consumer is frozen. This reads lag as evidence and ties it to the change, with the proof.
Stop reading lag as a number.
Name the consumer, the topic and the change behind it.