Menene liveness, readiness, da startup probes, kuma yaushe kake amfani da kowanne?

Question

IT Interview Practice · Accepted Answer

Probes suna bari kubelet ya bincika lafiyar container ya kuma yi aiki bisa sakamakon. Akwai uku: **liveness** (yana raye, ko ya kamata a sake kunna shi?), **readiness** (yana shirye ya karɓi zirga-zirga?), da **startup** (ya gama tashi tukuna?). Kowanne yana amsa tambaya daban.

## Probes uku a taƙaice

```text
LIVENESS   → "App ya makale/deadlock?"          KASA → kubelet YA SAKE KUNNA container
READINESS  → "Zai iya bautar zirga-zirga yanzu?" KASA → an cire Pod daga Service endpoints
                                                        (BA a sake kunna ba)
STARTUP    → "Tashin jinkirin ya gama?"          yana gudana NA FARKO; yayin da yake gudana,
                                                 an dakatar da liveness & readiness
```

## Readiness da liveness — mahimmancin bambanci

```text
Readiness yana sarrafa ZIRGA-ZIRGA:
  → an cire Pod da ya kasa readiness daga Service load-balancer,
    amma yana ci gaba da gudana (misali yana ɗumama cache ko dependency ya faɗi)
Liveness yana sarrafa RAYUWA:
  → an KASHE Pod da ya kasa liveness a kuma sake kunna shi (ya lalace fiye da gyara)
Amfani da liveness inda ka nufi readiness yana haifar da sake kunnawa marasa amfani.
```

## Bayanan probe

```yaml
spec:
  containers:
    - name: app
      image: myapp:1.0
      ports: [{ containerPort: 8080 }]

startupProbe:            # yana ba mai jinkirin tashi lokaci kafin sauran su fara
        httpGet: { path: /healthz, port: 8080 }
        failureThreshold: 30   # 30 * 10s = har zuwa mintuna 5 don tashi
        periodSeconds: 10

readinessProbe:          # shiryar da zirga-zirga
        httpGet: { path: /ready, port: 8080 }
        initialDelaySeconds: 5
        periodSeconds: 10

livenessProbe:           # sake kunnawa idan mara lafiya
        httpGet: { path: /healthz, port: 8080 }
        periodSeconds: 10
        failureThreshold: 3
```

## Hanyoyin probe

```text
httpGet    → GET a path; 2xx/3xx = wucewa       (mafi yawa ga web apps)
tcpSocket  → za mu iya buɗe haɗin TCP?           (ga service marasa HTTP)
exec       → gudanar da umarni; exit 0 = wucewa  (bincike na musamman)
grpc       → binciken lafiya na gRPC na asali
Manyan maɓallai: initialDelaySeconds, periodSeconds, timeoutSeconds,
           failureThreshold, successThreshold.
```

## Me ya sa startup probes suke wanzuwa

```text
App masu tashi a hankali (JVM warmup, manyan migrations) sun tilasta ka saita babban
initialDelaySeconds a kan liveness — wanda sannan ya sa ainihin faɗuwa ta yi jinkiri a gano.
startupProbe yana gyara wannan: yana kare tashin jinkiri, kuma KAWAI bayan ya wuce
ne liveness/readiness suke farawa da lokutansu na yau da kullum (matsattsu).
```

```bash
kubectl describe pod app     # duba kasawar probe & dalilan sake kunnawa a abubuwan da suka faru
kubectl get pod app          # rukunin READY yana nuna readiness; RESTARTS yana nuna liveness
```

## Me ya sa yake da muhimmanci

Health probes su ne yadda Kubernetes ke isar da self-healing da tuƙawar zirga-zirga cikin aminci, don haka fahimtar nau'ika uku da yaushe za a yi amfani da kowanne ilimi ne mai mahimmanci na aminci. Mahimmancin bambanci — wanda ke tafka wa injiniyoyi da yawa — shine tsakanin **readiness** da **liveness**: readiness yana sarrafa **zirga-zirga** (an cire Pod da ke kasawa daga Service load-balancer amma yana ci gaba da gudana, mafi kyau lokacin da yake ɗumama ko dependency ya faɗi na ɗan lokaci), yayin da liveness yana sarrafa **rayuwa** (an kashe Pod da ke kasawa a kuma sake kunna shi, don lokacin da ya makale fiye da gyara). Ruɗe su biyu — amfani da liveness probe inda ake nufin readiness — yana haifar da sake kunnawa marasa amfani da tsayawar aiki, don haka sanin wace tambaya kowane probe ke amsawa mai amfani ne kai tsaye. **Startup** probe yana warware ainihin ciwo na tarihi: app masu tashi a hankali sun taɓa tilasta babban `initialDelaySeconds` a kan liveness (wanda ya jinkirta gano ainihin faɗuwa), yayin da startup probe yana kare taga tashi ya kuma bari liveness/readiness su gudana da lokuta matsattsu, na yau da kullum kawai bayan tashi ya yi nasara. Fahimtar hanyoyin probe (`httpGet`, `tcpSocket`, `exec`, `grpc`) da maɓallan daidaitawa (`initialDelaySeconds`, `periodSeconds`, `failureThreshold`, da sauransu) yana bari ka saita binciken lafiya daidai maimakon da zato. Tunda probes suna yanke kai tsaye ko masu amfani suna bugun Pod masu lafiya kuma ko Pod da suka lalace suna farfaɗowa kai tsaye, samun su daidai na tushe ne don gudanar da workload masu aminci, na matakin production a kan Kubernetes.

Probes uku a taƙaice

Readiness da liveness — mahimmancin bambanci

Bayanan probe

Hanyoyin probe

Me ya sa startup probes suke wanzuwa

Me ya sa yake da muhimmanci

Tambayoyi masu alaƙa