What runs on it
Anything that should not block the request it was triggered by goes through a Redis-backed queue: an outgoing webhook, a transactional email, a Meilisearch index update, a KSeF submission. Each queue retries on failure and keeps a view of failed jobs in the admin panel.
Why queues matter for integrations
A KSeF outage or a slow webhook receiver on the integrator’s side does not stall order processing; the job sits in the queue and retries with back-off until it succeeds or a human intervenes. That is what keeps a temporary failure on one integration from becoming a platform-wide incident.
Redis itself holds only cache and queues, never a system of record: the catalog, orders and prices live in PostgreSQL, so flushing Redis never loses data, it only delays the jobs sitting in the queue at that moment.