Can we replay this work safely?
Separate delivery attempts from business effects, then verify the evidence and retention window.
Background Jobs Field Guide
Start with a concrete reliability question, then use the catalog when you need a specific model. Every resource is a bounded teaching tool or working reference—not a production configuration generator—and keeps your inputs local.
Choose the decision you need to make
Each path is short on purpose: establish the failure boundary, inspect one model, and leave with a checklist or operating decision.
Will the backlog drain by the deadline?
Estimate capacity, test controller lag, and define completion as a measurable objective.
Can old jobs survive a deployment?
Test payload compatibility, worker takeover, rollback, and replay before a mixed-version rollout.
Our jobs sometimes run twice
Separate progress from side effects, then examine the retry contract.
Our agent repeats actions after failure
Record intent, reconcile unknown results, and bound a replay.
Found an error or an assumption that needs a tighter boundary? Send a correction.
Filtering happens only in this page. Search text is not saved, added to the URL, or sent to analytics.
Tools grouped by problem
Size and scale
Capacity, admission, fairness, and the worker supply behind queue recovery.
- Backlog Drain-Time & Worker CapacityEstimate whether a queue can drain before your recovery deadline.Capacity calculator
- Tenant Fairness & BackpressureCompare tenant wait and rejection under competing scheduling policies.Scheduling simulator
- Queue Autoscaling PlaygroundSee whether worker startup and downstream limits defeat burst scaling.Scaling simulator
Retry and recover
Bound repeated work and decide what evidence is needed before recovery.
- Replay Readiness ChecklistReview the evidence needed before replaying failed work.Printable reference
- Retry Budget & Amplification WorkbenchBound nested attempts, jitter, and work that fits inside a deadline.Policy workbench
- Dead-Letter Queue TriageChoose the next evidence-gathering step for a failed job.Decision guide
- Kafka Recovery, Retry & DLQ PlannerIdentify the evidence and consistency boundary required before retry, quarantine, or replay.Recovery planner
Delivery and correctness
Protect one business intent across crashes, redelivery, and concurrent attempts.
- Async Failure LabSee how crashes, timeouts, and identity mistakes turn attempts into duplicate effects.Interactive lab
- Worker Ownership, Leases & FencingSee what happens when an expired worker wakes after takeover.Ownership explorer
- Transactional Outbox & Inbox Crash LabFollow state through commit, publish, effect, and acknowledgement crashes.Crash-point lab
- Kafka Delivery Contract ExplorerFind the exact boundary that still needs evidence before you rely on a delivery claim.Contract explorer
- Kafka Publishing Reliability LabSeparate leader append, replicated visibility, caller knowledge, and business identity.Publishing lab
- Kafka Transactions & Visibility LabSee why stored bytes, committed decisions, visible records, and caller knowledge can disagree.Transaction lab
- Kafka Consumer Progress & Ownership LabCommit the completed prefix without confusing group ownership with effect fencing.Consumer lab
- Kafka Effects, Outbox & Inbox WorkbenchChoose the identity and transaction that can actually protect one business operation.Effects workbench
Operate and evolve
Measure completion and change schedules or durable payloads without hiding risk.
- Scheduled Job & Overlap ExplorerPreview cron behavior through DST, missed runs, and overlap.Schedule explorer
- Job Completion SLO & Alert DesignerCount late and unfinished jobs in a deadline-based reliability objective.SLO workbench
- Job Payloads & Rolling DeploymentsTest old and new job shapes through rollout, rollback, delayed delivery, and replay.Compatibility lab