CKA Day 12 — Kubernetes DaemonSet, Job & CronJob Explained

Day 12 of the 40-day CKA full course by Tech Tutorials with Piyush. Duration: 19:30. Focus: three specialized workload controllers that handle node-local services, one-off tasks, and scheduled batch execution.

Synthesis

This video introduces three workload primitives beyond the standard Deployment: DaemonSet, Job, and CronJob. While Deployments are designed for stateless, horizontally scalable applications, these three controllers cover edge cases that are equally critical in production: running exactly one Pod per node, running a task to completion exactly once, and running a task on a schedule.

DaemonSet

A DaemonSet ensures that a copy of a Pod runs on every node (or a subset of nodes selected by a node selector). When a new node joins the cluster, the DaemonSet automatically schedules a Pod on it. When a node leaves, the Pod is garbage collected. Unlike a Deployment, which spreads replicas across nodes for high availability, a DaemonSet is one-to-one with nodes by design.

Canonical use cases:

• Node monitoring agents (Prometheus Node Exporter, Datadog Agent)
• Log collectors (Fluentd, Fluent Bit, Logstash) that need access to node-level logs
• CNI network plugins (Calico, Cilium, Flannel) that must run on every node to set up networking
• Storage daemons (Ceph OSD, Portworx) that manage local disk

YAML structure is nearly identical to a Deployment, but the kind is DaemonSet and apiVersion is apps/v1. There is no replicas field — the count is implicit (one per eligible node). The selector uses matchLabels just like a ReplicaSet.

Key difference from Deployment: A Deployment scales horizontally across the cluster; a DaemonSet scales vertically across the node axis. You cannot scale a DaemonSet with kubectl scale because the node count itself determines the replica count.

Practical commands:

kubectl get daemonset
kubectl describe daemonset <name>

Job

A Job creates one or more Pods and ensures that a specified number of them terminate successfully. It is designed for finite, batch-style workloads that run to completion rather than running indefinitely. When the Pod(s) finish successfully, the Job is marked Complete.

Key properties:

• completions: how many successful Pod completions are required (default: 1)
• parallelism: how many Pods can run concurrently (default: 1)
• backoffLimit: how many retries before marking the Job as failed (default: 6)
• activeDeadlineSeconds: maximum time a Job can run before Kubernetes terminates it

Use cases:

• Database migrations and schema updates
• One-off data processing or ETL tasks
• Batch report generation
• CI/CD pipeline steps that run inside the cluster

Restart policy: Jobs require restartPolicy: OnFailure or Never in the Pod template — Always is invalid because the Job controller must detect completion.

Practical commands:

kubectl get jobs
kubectl logs job/<name>  # reads logs from the job's pod
kubectl delete job <name>  # deletes the job and its pods

CronJob

A CronJob wraps a Job with a time-based schedule defined by a cron expression. It is the Kubernetes equivalent of crontab for cluster workloads. The CronJob controller creates a new Job object every time the schedule triggers, which in turn creates a Pod. This means failed executions do not block future scheduled runs because each run is an independent Job.

Key properties:

• schedule: standard cron syntax ("*/5 * * * *" for every 5 minutes)
• startingDeadlineSeconds: grace period for missed starts (e.g., if the controller was down)
• concurrencyPolicy: Allow (default), Forbid, or Replace — controls whether multiple Jobs from the same CronJob can run simultaneously
• successfulJobHistoryLimit / failedJobHistoryLimit: how many completed Job objects to retain (default 3 and 1)
• suspend: set to true to pause the schedule without deleting the CronJob

Use cases:

• Scheduled backups (database dumps, etcd snapshots)
• Periodic cleanup tasks (deleting old logs, pruning temp files)
• Recurring report generation (daily, weekly, monthly)
• Health-check probes that run on a timer rather than continuously

Exam tip: The CKA exam may ask you to create a CronJob with a specific schedule. Use kubectl create cronjob with --dry-run=client -o yaml to generate the manifest, then edit the schedule field.

kubectl create cronjob my-cron --image=busybox --schedule="*/5 * * * *" \
  --dry-run=client -o yaml > cronjob.yaml

Comparative Summary

Controller	Lifetime	Scaling Axis	Use Case	apiVersion
Deployment	Indefinite	Horizontal (replicas)	Long-running stateless apps	apps/v1
DaemonSet	Indefinite	Node count (one per node)	Node-local agents, monitoring, networking	apps/v1
Job	Finite (to completion)	Parallelism + completions	One-off batch tasks, migrations	batch/v1
CronJob	Scheduled (finite per run)	Node + time axis	Recurring backups, cleanup, reports	batch/v1

CKA Exam Relevance

• Workloads & Scheduling (~15%): Expect tasks to create a DaemonSet, a Job, or a CronJob from a given spec.
• Troubleshooting (~30%): Debug why a Job is stuck (check restartPolicy, backoffLimit, image pull errors) or why a CronJob missed a schedule (check startingDeadlineSeconds and controller health).
• Speed patterns:
  • kubectl create job ... --dry-run=client -o yaml for quick Job manifests
  • kubectl create cronjob ... --dry-run=client -o yaml for CronJob templates
  • Remember that restartPolicy: Always is invalid inside a Job template

See Also

Wiki Concepts

• Kubernetes DaemonSet — node-level workload controller deep dive
• Kubernetes Jobs — batch execution and completion semantics
• Kubernetes CronJobs — scheduled batch workloads
• Pod Fundamentals — the unit all these controllers manage
• Deployment, ReplicaSet & Replication Controller — the standard long-running workload controller
• Kubernetes Architecture — where kube-proxy and CoreDNS run as DaemonSets/Deployments
• CKA Certification — exam structure and domains
• CKA Study Roadmap — 40-day schedule

Related Sources

• CKA Day 8 — Deployments, ReplicaSets & Replication Controllers — the foundation for workload controllers
• CKA Day 7 — Pod Explained — Pod fundamentals

Creator / Entity

• Tech Tutorials with Piyush — CKA 40-day course creator

---

Tags: kubernetes daemonset job cronjob cka workload batch scheduling devops