Kif jaħdmu l-circuit breaker u r-retries b'backoff f'sistemi distribwiti?

Question

Accepted Answer

F'sistemi distribwiti, l-ċills jifšlu. **Ir-retries b'backoff** jirreküperaw minn żbalji *temporanji*; **circuit breaker** iwaqqfek milli tħebbeż dipendenza li verament *down*. Huma komplemantarji: erġa' ħaġċa l-ħajwa, aqbad fuq l-outage.

## Retry b'exponential backoff + jitter

Għal difetti temporarji (timeout, problemi network qosra, 503), erġa' ħaġċa — imma **backoff exponenzjali u żid jitter** biex ir-retries jinqsmu minflok jinqabdu f'gregge tandil.

```text
attempt 1 → wait ~1s   (+ random jitter)
attempt 2 → wait ~2s   (+ random jitter)
attempt 3 → wait ~4s   (+ random jitter)
→ cap at maxRetries (e.g. 3) and a max delay → don't retry forever
```

Avviż kritiċi:

- **Kap ir-retries** — retries bla limitu jigganaw **retry storm** li toverload servizz li jirkupera.
- **Idempotency** — ħaġċa biss operazzjonijiet li huma sikuri li jerġaw (pagament riħaġċa m'għandux iċċarġi darbtejn).
- **Jitter** — mingħajr tiegħu, il-klients kollha erġaw ħaġċa nifs-sekonda u terġa' tuża l-load.

## Circuit breaker

Circuit breaker itrackja l-fardel għal dipendenza u, wara soglia, **infetaħ** — jifšal mgliegħ (jew jirritorna fallback) minflok jħadd servizz mejjet. Wara cooldown jinqala' **semi-open** biex jinvestiga r-rikupru.

```text
CLOSED   → calls pass through; count failures
  too many failures → trip → OPEN
OPEN     → fail fast immediately (no call); start cooldown timer
  cooldown elapsed → HALF-OPEN
HALF-OPEN → allow a few probe calls
  success → CLOSED (recovered) ; failure → back to OPEN
```

## Meta kull wieħed japplikaw

```text
Retry           → transient, likely-to-succeed-soon errors (1 slow call)
Circuit breaker → repeated/sustained failures (the dependency is down)
→ use together: breaker prevents retries from piling onto a dead service
```

## Għaliex ta' importanza

Bingħajr dawn, dipendenza ta' falliment waħda tieħu l-caller għall-down: ir-requests jakkumulaw fuq timeout, ir-retries jamplifikaw il-load, u l-falliment jinkaskad fuq servizzi. Exponential backoff b'jitter jirkupera minn ħajwa mingħajr retry storm; kapping u idempotency iżommu r-retries sikuri; u circuit breaker isseħħ idda mal-ħamrija ta' risorsi fuq dipendenza mejja u tagħtihom spazju biex jirkupraw. Flimkien jbiddlu falliment ta' dipendenza f'avveniment kontnut u auto-healing.