12.2Why duplicates happen even when your code is “correct”

Imagine:

Mule
  ↓
Salesforce Upsert
  ↓
Salesforce commits successfully
  ↓
response gets lost
  X
Mule sees timeout

Mule cannot know whether Salesforce committed.

So caller retries:

same request again

This is normal distributed-system behavior.

That is why this sentence is worth memorizing:

“A timeout is an ambiguous outcome, not proof that the downstream write failed.”