Lag: Slow Consumer or Producer Spike?Kafka · medium · ~12 minThe notifications consumer group's lag has grown from ~0 to 9 million over three hours and is still climbing. On-call is split: half say 'the consumers are broken, restart them', half say 'we're just getting more traffic'.The metrics settle it — and dictate completely different responses. Which is it?Evidence

Metrics

Config

Checkpoints

Slow consumer or producer spike — what do the metrics say?

Why is 'restart the consumers' the wrong reflex here?

Each consumer sustains ~5k msg/s and input is steady at ~46k msg/s. What is the MINIMUM number of consumers that stops lag from growing? (whole number)

You scale to that number — but is there a ceiling in the topic itself? What would you tell the team about longer-term capacity? (interview-style)