Queued but Never RunningPipeline · easy · ~10 minSince about 07:30, dozens of tasks across many DAGs sit in queued and never start. The scheduler is healthy, workers are healthy, CPU is nearly idle. Meanwhile a handful of tasks have been 'running' for 5+ hours — a new wait_for_partner_file sensor family deployed yesterday, and two API-poll tasks that loop internally.Nothing is broken, and nothing is running. Explain that sentence, then fix it.Evidence

Metrics

Evidence

Config

Checkpoints

Why is an idle cluster refusing to run anything?

The durable fix for the waiters?

Why is 'just add slots' the losing move, and how do you narrate the fix? Interview-style.