Argo Workflows & Argo Events
The other half of the Argo Project. Learn the container-native workflow engine where every step is a Pod — templates, artifacts, DAGs and fan-out — and the EventSource, EventBus and Sensor chain that starts it all from an outside event.
- flashcards
- 90
- flashcards
- per day
- ~10 min
- per day
- level
- Intermediate
- level
- modules
- 6
- modules
What are Argo Workflows and Argo Events?
Argo Workflows is a container-native workflow engine that runs on Kubernetes as a CRD. You describe a pipeline as a Workflow, whose spec.entrypoint names the template that acts as its main function, and every Step or DAG Task the controller reaches becomes its own Pod. That single fact explains most of the engine's behaviour: why data between steps travels as parameters or as artifacts through an object store rather than on a shared disk, why an init container stages inputs and a wait container saves outputs, and why the cost of a workflow scales with how many steps it has.
Templates are the unit of work and behave like functions. A container template runs your image, script wraps one and captures its stdout as outputs.result, resource acts on cluster objects through the API, suspend holds the run for a duration or a human approval, and containerSet puts several containers in one Pod. steps and dag are not work at all — they are invocators that call other templates, which is the distinction most newcomers get wrong first.
Reuse and scheduling live one level up: a WorkflowTemplate is a definition stored in the cluster, ClusterWorkflowTemplate is its cluster-scoped twin, and a CronWorkflow fires a workflow spec on a schedule with its own concurrency policy and history limits. For fan-out, withItems takes a list you write, withParam takes a JSON array a previous step printed at runtime, and withSequence counts — so a pipeline can decide how wide it should be while it is running.
Argo Events is the inverse arrow. An EventSource knows how to consume from more than twenty external systems — webhooks, S3, Kafka, GitHub, Pub/Sub — normalises what arrives into CloudEvents and publishes it to the EventBus. A Sensor subscribes, holds a set of event dependencies, filters them, and executes triggers once its conditions resolve. The two halves never talk to each other directly, which is what lets several Sensors react to one event and one Sensor combine events from different sources.
This deck is deliberately separate from the GitOps track. Argo CD and Argo Rollouts are continuous delivery; a DAG engine and an event router are a different subject that happens to share a name. Together the two tracks cover the whole Certified Argo Project Associate blueprint — this one is its largest domain, Argo Workflows at 36 %, plus Argo Events at 12 %.
6 modules, seed to bloom
Each module is a set of flashcards — 90 in total. Answer, review, and watch your knowledge grow from seed to full bloom.
Workflow Fundamentals
The Workflow resource, entrypoint and templates, template types, and the controller, server and pod anatomy
15 cardsThe Workflow Spec
Passing parameters, sequencing with steps, conditionals, retry strategies, exit handlers, lifecycle hooks and cleanup
15 cardsTemplates & Reuse
WorkflowTemplates and ClusterWorkflowTemplates, template references, CronWorkflows, memoization and synchronization
15 cardsArtifacts
Producing and consuming artifacts, archive strategies, repository configuration and precedence, and artifact garbage collection
15 cardsDAGs & Data Processing
DAG dependencies and enhanced depends logic, fail-fast behaviour, loops with withItems, withParam and withSequence, and dynamic fan-out
15 cardsArgo Events
EventSources, the EventBus, Sensors and their dependencies, trigger types and operations, conditions, filters and parameterization
15 cardsSample questions
A taste of the real flashcards. Pick an answer, then reveal the explanation.
Which template type schedules a container running an image you name?
- Acontainer — it schedules a single container, the most common template type
- Bscript — it wraps a container and adds an inline source body to execute
- Cresource — it performs get, create, apply or delete on cluster resources
- Dsuspend — it pauses execution for a duration or until someone resumes it
What input does withParam expect?
- AA JSON array of items, which may be generated at runtime
- BA YAML list written inline in the template that uses it
- CA comma-separated string the controller splits on commas
- DA single parameter whose value the loop repeats N times
How are output artifacts packaged by default?
- AAs a tarball that is then gzipped before it is uploaded
- BAs the raw file, uploaded exactly as the container wrote it
- CAs a zip archive, because object stores index those cheaply
- DAs a gzipped tarball for directories, with single files sent raw
Which concurrencyPolicy removes older Workflows before scheduling a new one?
- AReplace — remove all old runs before the new one is scheduled
- BForbid — do not allow any new run while an old one is going
- CAllow — let the scheduled runs proceed alongside each other
- DSuspend — hold the schedule until an operator releases it again
What role does the EventBus play?
- AThe transport layer — sources publish to it and sensors subscribe
- BThe filter layer — it discards events that no sensor is waiting for
- CThe trigger layer — it creates the workload once conditions hold
- DThe ingress layer — it terminates the webhooks that sources expose
Learn it once, keep it for good
Answer a question
Each card is one practical concept with multiple options. Pick what you think is right.
Get the full answer
See the correct option plus a clear explanation, and a link to deeper docs when one is available.
Review at the right time
A spaced-repetition engine (SM-2 or FSRS) resurfaces each card just before you would forget it.
Why Argo Workflows is worth your time
Batch and data pipelines that live in the cluster
ETL, model training, CI-style jobs and scheduled reports run as Kubernetes-native DAGs instead of a separate scheduler.
Fan-out you can reason about
withParam, withItems and withSequence turn one step into hundreds, and depends logic says exactly what waits on what.
Event-driven without glue code
A webhook, a queue message or an object landing in a bucket can start a workflow with no service of your own in between.
The biggest domain on the CAPA exam
Argo Workflows is 36 % of the Certified Argo Project Associate paper and Argo Events another 12 % — nearly half of it.
Common questions
Do I need to know Kubernetes first? +
Yes. Every workflow step becomes a Pod, artifacts move through object storage, and templates are fields on a custom resource, so Pods, ConfigMaps and Secrets should already be familiar. The Kubernetes track pairs well with this one.
How is Argo Workflows different from Argo CD? +
They share a name and a controller pattern and nothing else. Argo CD reconciles a cluster to declared state in Git — that is continuous delivery, covered by the GitOps track. Argo Workflows runs pipelines: DAGs, batch jobs and data processing, one Pod per step.
Is this the same as Argo Events on its own? +
Argo Events is one of the six modules here. It covers EventSources, the EventBus and its NATS and Kafka implementations, Sensors and their dependencies, trigger types and operations, conditions, filters and trigger parameterization.
Does it cover Argo Workflows for CI? +
The mechanics it needs, yes — templates, artifacts, retries, exit handlers and DAG dependencies are the same whether the pipeline builds software or processes data. Whether Workflows is the right CI tool for you is a design question this deck does not argue.
Is it free? +
Yes, completely free. No registration or credit card is required, and all your progress is stored locally in your browser.
How long does it take? +
About 10 minutes a day. Spaced repetition means short, frequent sessions beat long cramming — most learners get comfortable with the workflow spec within a couple of weeks.
Ready to master Argo Workflows?
Plant your first seed today. Ten minutes a day is all it takes to make the workflow spec second nature.