X'inhi d-differenza bejn CPU/memory requests u limits, u kif jaffettwaw l-iscceduling, l-eviction, u l-QoS?

Question

IT Interview Practice · Accepted Answer

**Requests** huma dak li container huwa *garantit* (użat għall-iscceduling); **limits** huma s-*saqaf* li ma jistax jaqbeż (infurzat fir-runtime). Flimkien imexxu fejn jitqiegħdu l-Pods, kif jiġu throttled jew maqtula taħt pressjoni, u liema klassi ta' **Quality of Service (QoS)** jieħu Pod.

## Requests vs limits

```text
REQUEST → the amount RESERVED for the container. The SCHEDULER uses the sum of
          requests to decide if a node has room. This is your guaranteed floor.
LIMIT   → the MAX the container may use. Enforced at runtime:
          CPU over limit    → THROTTLED (slowed down, not killed)
          memory over limit → OOM-KILLED (container terminated, then restarted)
```

```yaml
spec:
  containers:
    - name: app
      image: myapp:1.0
      resources:
        requests:
          cpu: "250m"        # 0.25 ta' core, garantit
          memory: "256Mi"
        limits:
          cpu: "500m"        # jista' jaqbeż sa 0.5 core, imbagħad throttled
          memory: "512Mi"    # jaqbeż dan → OOMKilled
```

## Kif l-iscceduling juża r-requests

```text
Node has 4 CPU / 8Gi allocatable.
Scheduler sums the REQUESTS of Pods already placed; a new Pod fits only if its
requests still fit in what remains. (Limits are NOT used for placement.)
→ Set requests too low → node overcommitted, Pods fight for resources.
→ Set requests too high → wasted capacity, Pods stuck Pending (nothing fits).
```

## Klassijiet QoS (jiddeċiedu l-ordni tal-eviction taħt pressjoni tal-memorja)

```text
Guaranteed  → requests == limits for BOTH cpu & memory (every container)
              → last to be evicted (highest priority)
Burstable   → has requests but they are < limits (or partially set)
              → evicted after BestEffort
BestEffort  → no requests or limits at all
              → FIRST to be evicted when the node runs low on memory
```

```bash
kubectl describe pod app | grep -i qos    # QoS Class: Guaranteed/Burstable/BestEffort
kubectl top pods                           # użu attwali tas-CPU/memorja
kubectl describe node <node>               # riżorsi allokati vs kapaċità
```

## Gwida prattika

```text
✓ Always set memory requests = limits for predictable, OOM-safe behavior on critical apps
✓ Set CPU requests to the steady-state need; CPU limits are often left off to allow bursting
✓ Right-size from real usage (kubectl top / metrics) — not guesses
✓ Use LimitRange / ResourceQuota to enforce sane defaults per namespace
```

## Għaliex huwa importanti

Ir-requests u l-limits huma kif Kubernetes jaqsam ir-riżorsi finiti tan-node b'mod ġust u prevedibbli, għalhekk li tifhem id-distinzjoni u l-konsegwenzi tagħha huwa għarfien essenzjali tal-prestazzjoni u l-affidabbiltà. L-għarfien ewlieni huwa li jservu skopijiet differenti: ir-**requests** huma riservazzjoni garantita li l-**scheduler** juża biex iqiegħed il-Pods (jiġbor ir-requests biex jiddeċiedi jekk node għandux spazju, u jinjora l-limits), filwaqt li l-**limits** huma saqaf tar-runtime infurzat b'mod differenti għal kull riżorsa — li taqbeż limit tas-**CPU** iġiblek **throttled** (imnaqqas), imma li taqbeż limit tal-**memorja** iġiegħel il-container jiġi **OOM-killed** u jerġa' jibda. Li tqassam ir-requests ħażin għandu spejjeż reali fuq iż-żewġ naħat: baxx wisq u n-nodes isiru overcommitted u l-Pods iġewħu lil xulxin; għoli wisq u l-kapaċità tinħela filwaqt li l-Pods joqogħdu Pending għax xejn ma jidħol. It-tliet **klassijiet QoS** — Guaranteed (requests ugwali għal-limits), Burstable (requests taħt il-limits), u BestEffort (xejn issettjat) — jiddeterminaw l-ordni tal-eviction taħt pressjoni tal-memorja, bil-BestEffort jinqatel l-ewwel u l-Guaranteed l-aħħar, li huwa eżatt dak li teħtieġ tifhem biex tipproteġi workloads kritiċi. Li tkun taf il-gwida prattika (issettja memory requests ugwali għal-limits għal apps kritiċi, tqassam id-daqs sew mill-użu reali permezz ta' `kubectl top`, tinforza defaults b'LimitRange/ResourceQuota) tibdel dan minn teorija għal ippjanar tajjeb tal-kapaċità. Peress li l-konfigurazzjoni ħażina tar-riżorsi hija waħda mill-aktar kawżi komuni ta' instabbiltà, nefqa moħlija, u Pods Pending fi clusters reali, li tqabbad ir-requests, il-limits, u l-QoS huwa fundamentali biex tħaddem workloads ta' Kubernetes effiċjenti u stabbli.

Requests vs limits

Kif l-iscceduling juża r-requests

Klassijiet QoS (jiddeċiedu l-ordni tal-eviction taħt pressjoni tal-memorja)

Gwida prattika

Għaliex huwa importanti

Mistoqsijiet relatati