Introduction

NexJob is a reliable, open-source background job processing library for .NET 8. It was built for developers who need jobs that must run, fail safely, and leave a trace — without paying for storage providers, wrestling with plugin ecosystems, or accepting silent failures. NexJob gives you predictable execution, built-in retries, deadline enforcement, dead-letter handling, and a real-time operational dashboard as first-class features from day one.

Why NexJob?

If you have used Hangfire, you know the pain: free storage only covers InMemory, deadlines require a plugin, and observability is an afterthought. NexJob fixes all of that.

Feature	NexJob	Hangfire
Storage providers	5 free (PostgreSQL, SQL Server, Redis, MongoDB, InMemory)	Free InMemory only; others require a paid license
Deadline enforcement	Built-in (deadlineAfter)	Plugin required
Dead-letter handling	Automatic after exhausted retries	Manual
Dispatch latency	Near-zero (wake-up channel)	Polling-based
Dashboard	Built-in, standalone UI	Built-in (Pro required for advanced features)
OpenTelemetry	Built-in traces and metrics	Plugin required
Concurrency throttling	[Throttle] attribute per resource	Queue-level limits only
Ecosystem	Young library, focused scope	Mature ecosystem, many plugins
Package size	~50 KB	~2 MB

NexJob is not a drop-in replacement for Hangfire. If you need calendar-based scheduling, distributed execution across untrusted networks, or a mature enterprise plugin ecosystem, Hangfire is the better fit. NexJob focuses on a tighter scope and delivers it with exceptional performance.

Benchmark numbers

All measurements are per individual enqueue operation on .NET 8 (BenchmarkDotNet v0.14, RyuJIT AVX2, in-memory storage baseline):

Metric	NexJob	Hangfire
Latency (mean)	13.35 µs	35.95 µs — 2.7× slower
Memory allocated	2.10 KB	11.20 KB — 81% more
GC Gen0 (per 1 k ops)	0.06	0.85 — 14× more collections
GC Gen1 (per 1 k ops)	0.00	0.18 — non-zero Gen1 cost

Hangfire's overhead comes from runtime LINQ expression-tree parsing and reflection during enqueue. NexJob avoids both.

Packages

Install only what you need. The core package ships with InMemory storage and the full dispatcher — every other package is optional.

Package	Description
NexJob	Core library: dispatcher, scheduler, InMemory storage, retries, deadlines
NexJob.Postgres	PostgreSQL storage provider
NexJob.SqlServer	SQL Server storage provider
NexJob.Redis	Redis storage provider
NexJob.MongoDB	MongoDB storage provider
NexJob.Dashboard	Embedded ASP.NET Core dashboard middleware
NexJob.Dashboard.Standalone	Embedded HTTP dashboard server for Worker Services
NexJob.OpenTelemetry	OpenTelemetry SDK instrumentation (traces and metrics)
NexJob.Trigger.AzureServiceBus	Azure Service Bus trigger
NexJob.Trigger.AwsSqs	AWS SQS trigger
NexJob.RabbitMQ	RabbitMQ trigger and resilient outbox producer
NexJob.Kafka	Apache Kafka trigger and resilient outbox producer
NexJob.Trigger.GooglePubSub	Google Cloud Pub/Sub trigger
NexJob.Trigger.Salesforce	Salesforce Pub/Sub API trigger (gRPC and Avro)
NexJob.Trigger.SalesforceStreaming	Salesforce Streaming API trigger (CometD and Bayeux)

Key features

Write and run jobs

• IJob / IJob<T> — clean interfaces for parameterless and typed jobs, with full dependency injection support
• Job filters — implement IJobExecutionFilter for cross-cutting middleware (logging, correlation IDs, etc.)
• Job context, progress and checkpoints — report progress, and save a checkpoint so a retried job resumes where it stopped instead of starting over
• Job continuations — chain jobs with parent/child relationships using ContinueWithAsync
• Job priority — JobPriority controls execution order within a queue
• Tags and idempotency keys — label jobs and find them again

Stay reliable

• Predictable retries — configurable global delay policy plus per-job [Retry] attribute with exponential backoff
• Failures that should not be retried — list exception types such as ArgumentException and the job goes straight to dead-letter instead of burning attempts
• Per-job attempt limit — pass maxAttempts when enqueuing, so the same job type can fail fast for one caller and retry longer for another
• Execution timeout — [ExecutionTimeout] cancels a job's token after a limit and treats it as a normal failure (opt-in, cooperative)
• Dead-letter handlers and forwarders — an IDeadLetterHandler<T> runs when all retries are exhausted, and an IDeadLetterForwarder sees every such job, which is the hook for alerts
• Crash recovery — a job left behind by a node that died is found by its stale heartbeat and run again, or dead-lettered if it had no attempts left
• Deadline enforcement — jobs expire before execution if deadlineAfter has elapsed; no zombie jobs
• Idempotency — DuplicatePolicy controls re-enqueue behavior for jobs with the same idempotency key
• Queue isolation by default — the default queue is {prefix}.default per application, so services that share a database do not share a queue; the legacy default is still drained
• Queue circuit breaker — pauses a queue automatically when a dependency is down, probes it with one job, and ramps back up gradually
• Concurrency throttling — [Throttle] caps per-resource concurrency locally; AddNexJobDistributedThrottle() enforces cluster-wide limits via Redis
• Execution windows — restrict a queue to certain hours and days, such as nights only or business days
• Graceful shutdown — running jobs are given time to finish, and the ones interrupted go back to the queue without using an attempt
• Delivery guarantees — one table of what each failure costs a job

Schedule

• Recurring jobs — register in code with RecurringAsync or declare them in appsettings.json with full timezone support
• Delayed and scheduled jobs — run after a delay or at a specific time

Operate

• Built-in dashboard — standalone dark UI with live cluster topology, SSE log stream, and job catalog
• Dashboard authorization — implement IDashboardAuthorizationHandler to control who can view or operate the dashboard
• Runtime control — pause and resume queues, requeue failed jobs, delete jobs and reset circuit breakers at runtime with IJobControlService
• Alerts — know when a job fails for good, with a Slack recipe and the metrics to alert on
• Health checks — AddNexJob() on the ASP.NET Core health checks builder reports whether storage is reachable
• OpenTelemetry — distributed traces and metrics emitted out of the box
• Job retention — automatic cleanup of terminal jobs with configurable TTL via the [Retention] attribute
• Configuration — NexJobOptions centralises worker count, polling interval, retry policy, and more

Storage and integrations

• Five storage providers — PostgreSQL, SQL Server, Redis, MongoDB and in-memory, with no paid tier
• Read replicas — UseDashboardReadReplica() offloads dashboard queries to PostgreSQL or SQL Server read replicas
• Several services on one database — queue isolation, and foreign jobs from other microservices are deferred without penalizing attempt counts
• External triggers — turn messages from Kafka, RabbitMQ, AWS SQS, Azure Service Bus, Google Pub/Sub and Salesforce into jobs
• Resilient Outbox — transaction-safe event producers for RabbitMQ and Apache Kafka

Trust

• Tested on real databases — reliability scenarios with several nodes, a killed process, upgrades from the previous version, and an honest list of what is not covered

Note

NexJob is released under the MIT License and is free for commercial use with no storage-provider paywalls or enterprise tiers. See the LICENSE file for full terms.

Where to go next

• Quick Start — install NexJob and run your first job in under 5 minutes
• Mental Model — understand storage-first design, the job state machine, and crash recovery before writing production jobs