Streams & Workers
Events in, work done,
and you can see all of it.
Some data arrives as a torrent, and some work takes an hour. Quadrazene runs durable consumers for the first and a worker framework for the second, with the same visibility, permissions and audit as everything an operator does by hand.
Source
A Redis stream or Kafka topic, named by a Connection.
Consume
A long-lived, self-reconnecting consumer in a consumer group.
Route
Fire a Formula per message, or batch-upsert straight into a table.
Protect
Retry, then dead-letter a poison message to a ticket.
Watch
Ingested count, last seen and lag on a live status page.
Durable by design
A consumer that survives restarts and bad messages.
Redis Streams
Consumer-group reads with acknowledgement, so a restart resumes from the group's last acknowledged position. There is no separate checkpoint table to lose.
Kafka
Native consumer-group offsets, committed only after a message is handled or dead-lettered. At-least-once, never silently skipped.
Many streams, one binding
A single binding can read several stream keys together, with TLS and Sentinel supported for managed Redis.
Poison-message safe
A message that keeps failing is retried, then dead-lettered to a Resolver ticket and acknowledged, so one bad record cannot block the stream. It works even with read-only credentials.
Two sinks
Fire a Formula per message with a filter and input template, or batch-upsert straight into a table without touching the reaction engine. Do both when you need to.
Any broker in the catalog
Other brokers, such as Event Hubs, Kinesis, SQS, SNS, RabbitMQ and NATS, are in the connector catalog for use as Connections and event trigger sources.
Long-running work
Jobs are runs, so they inherit everything.
One worker framework
A background job is an ordinary run. Workers claim one atomically, report progress by phase, and finish with structured output or a recorded failure.
Graceful shutdown
Handlers receive an abort signal, so a deploy stops work at a safe point instead of orphaning it.
Engine console
Per-tenant engine status, the live queue, routing rules and schedules, with a master switch for scheduling.
Installation-wide view
Platform administrators see every engine's heartbeat and job counts across all tenants in one screen.
Consumer health
Stream consumers report heartbeats too: events ingested, last seen per stream, dead-letter count, last error, and consumer-group lag.
Radar-ready
Put ingest lag and consumer health on a Radar dashboard so the pipeline is watched like any other service.
Point it at your stream.
Give us a topic or a stream key and we'll show events landing, a poison message being caught, and the consumer on the status page.