Running work outside requests.
WHAT EVERY STACK NEEDS
A queue Worker processes consuming it A scheduler for periodic work
WHAT THE QUEUE HOLDS
Jobs: a type, a payload, and metadata.
WHAT TO KEEP SMALL
The payload.
WHY
Large payloads slow the queue; pass identifiers and fetch the data.
WHAT TO CONFIGURE
Retry count Increasing delay between attempts A destination for permanently failed jobs
WHAT TO ENSURE
That processing twice is safe.
HOW
Record completion, and check before acting.
WHAT TO MONITOR
Queue depth Failure rate Whether workers are alive Time jobs wait before starting
WHAT A GROWING QUEUE INDICATES
Insufficient workers, or jobs failing repeatedly.
WHAT TO SEPARATE
Queues by priority, so slow bulk work does not delay urgent work.
WHAT TO BE CAREFUL WITH
Jobs holding database connections for a long time Jobs with no timeout Workers running old code after deployment
WHAT TO ALERT ON
Failed jobs, and a job type that stops appearing.
