Kaj so StatefulSets, PersistentVolumes/Claims in kako poganjate delovne obremenitve s stanjem na Kubernetesu?

Question

IT Interview Practice · Accepted Answer

Delovne obremenitve s stanjem (podatkovne baze, čakalne vrste, gručaste shrambe) potrebujejo **stabilno identiteto** in **trajno shrambo** — stvari, ki jih Deployment ne zagotavlja. **StatefulSet** Podom da stabilna imena in urejen življenjski cikel; **PersistentVolumes (PV)** in **PersistentVolumeClaims (PVC)** pa jim dajo shrambo, ki preživi Pod.

## Zakaj Deployments ne zadostujejo

```text
Deployment Pods are INTERCHANGEABLE: random names (web-7f9-abc), any order, share
nothing per-Pod. That's perfect for stateless apps — but a database replica needs:
  → a STABLE network identity (so peers can find it: db-0, db-1, db-2)
  → its OWN durable disk that survives restarts/rescheduling
  → ORDERED, controlled startup/scaling (bring up db-0 before db-1)
StatefulSet provides exactly these.
```

## Jamstva StatefulSet

```text
Stable NAMES     → pod-0, pod-1, pod-2 (ordinal, not random)
Stable DNS       → db-0.db.default.svc... via a headless Service
Stable STORAGE   → each Pod gets its OWN PVC (db-0 keeps its disk across restarts)
Ordered ops      → create/scale up in order 0,1,2…; terminate in reverse
```

## Trajna shramba: PV, PVC, StorageClass

```text
StorageClass → the "how to provision" template (e.g. AWS gp3, fast-ssd)
PVC          → a REQUEST for storage ("I need 20Gi, ReadWriteOnce")   ← app-facing
PV           → the ACTUAL provisioned volume bound to the PVC          ← cluster-facing
Dynamic provisioning: a PVC referencing a StorageClass auto-creates a matching PV.

App → PVC → (StorageClass provisions) → PV → real disk (EBS, PD, Ceph, NFS…)
```

## StatefulSet s shrambo po Podu

```yaml
apiVersion: apps/v1
kind: StatefulSet
metadata:
  name: db
spec:
  serviceName: db            # headless Service za stabilen DNS
  replicas: 3
  selector:
    matchLabels: { app: db }
  template:
    metadata:
      labels: { app: db }
    spec:
      containers:
        - name: db
          image: postgres:16
          ports: [{ containerPort: 5432 }]
          volumeMounts:
            - name: data
              mountPath: /var/lib/postgresql/data
  volumeClaimTemplates:      # vsak Pod dobi SVOJ PVC/PV iz te predloge
    - metadata:
        name: data
      spec:
        accessModes: [ReadWriteOnce]
        storageClassName: fast-ssd
        resources:
          requests: { storage: 20Gi }
```

```yaml
# Headless Service, ki da stabilen DNS po Podu (db-0.db, db-1.db, …)
apiVersion: v1
kind: Service
metadata: { name: db }
spec:
  clusterIP: None            # headless
  selector: { app: db }
  ports: [{ port: 5432 }]
```

## Načini dostopa in opozorila

```text
ReadWriteOnce (RWO)  → mounted read-write by ONE node (most block storage: EBS/PD)
ReadOnlyMany (ROX)   → read-only by many nodes
ReadWriteMany (RWX)  → read-write by many nodes (needs NFS/CephFS/EFS)
⚠ Deleting a StatefulSet does NOT delete its PVCs by default (data is preserved).
⚠ Reclaim policy (Retain/Delete) decides what happens to the PV when the PVC is removed.
⚠ Kubernetes gives identity+storage, but app-level clustering/replication is YOUR job
  (often via an Operator, e.g. for Postgres/Kafka).
```

```bash
kubectl get statefulset,pvc,pv
kubectl get pods -l app=db      # db-0, db-1, db-2 (stabilni, urejeni)
```

## Zakaj je to pomembno

Pravilno poganjanje delovnih obremenitev s stanjem je eden težjih delov Kubernetesa, zato je razumevanje StatefulSetov in trajne shrambe znanje senior ravni z resničnimi produkcijskimi posledicami. Jedrni uvid je *zakaj* Deployments zaostajajo: njihovi Podi so zamenljivi (naključna imena, poljuben vrstni red, brez stanja po Podu), kar je idealno za storitve brez stanja, a napačno za podatkovne baze in gručaste shrambe, ki potrebujejo **stabilno identiteto** (da peeri prek DNS najdejo `db-0`, `db-1`, `db-2`), **svoj trajni disk**, ki preživi ponovne zagone in prerazporejanje, ter **urejen, nadzorovan** zagon in skaliranje. **StatefulSet** ponuja natanko ta jamstva in se povezuje z abstrakcijami shrambe: **PersistentVolumeClaim** je zahtevek po shrambi s strani aplikacije, **PersistentVolume** je dejansko dodeljeni nosilec, **StorageClass** pa omogoča dinamično dodeljevanje — pri čemer `volumeClaimTemplates` vsakemu Podu da svoj PVC/PV. Opozorila so tam, kjer si senior inženirji prislužijo svoj kruh: **načini dostopa** (RWO proti RWX in kateri zaledji jih podpirajo), dejstvo, da izbris StatefulSeta privzeto **ohrani njegove PVC-je**, **reclaim policy**, ki ureja usodo PV, in — kritično — da Kubernetes zagotovi identiteto in shrambo, a **replikacija in gručenje na ravni aplikacije ostaneta vaša odgovornost**, pogosto prepuščena Operatorju. Ker sta trajnost podatkov in pravilno vedenje s stanjem neusmiljena (napake pomenijo izgubo podatkov ali split-brain), je obvladovanje StatefulSetov, PV/PVC in njihovih opozoril temeljno za zanesljivo poganjanje podatkovnih baz in drugih sistemov s stanjem na Kubernetesu.

Zakaj Deployments ne zadostujejo

Jamstva StatefulSet

Trajna shramba: PV, PVC, StorageClass

StatefulSet s shrambo po Podu

Načini dostopa in opozorila

Zakaj je to pomembno

Povezana vprašanja