Ninety-Nine Tasks Done in Four Minutes, One in Fifty
Read what each party saw, commit to a cause, and only then find out which of them was right. The root cause, why the obvious reading was wrong, and the fixes are all held back until you have answered.
The symptom
What was reported, before anyone knew what was happening.
What each party saw, and what each concluded
The evidence, in the form it actually arrived in — several parties, several partial views, several confident conclusions.
| Who | What they could actually see | What they concluded | Verdict |
|---|---|---|---|
| Scheduler | 512 tasks dispatched; 511 complete within 6 minutes; one still running. | The job is 99.8% complete; wait for the last task. | ✓ right |
| The slow task’s worker | A partition 60× larger than the median, processing at normal per-record speed. | I am healthy and making progress. | ✓ right |
| Capacity engineer | 90% idle cluster during a job that is over its window. | The job is under-parallelised; increase the worker count. | ✕ wrong |
| Data engineer | Total input volume up only 8% year on year. | Growth cannot explain a 4.5× runtime increase. | ✓ right |
1 of 4 parties reasoned correctly from what they could see and still reached the wrong conclusion. Nobody in this table is careless. Each one acted on complete-looking local information, and the information was local. That gap — between what a node can observe and what is true — is the whole domain, and one of these readings will usually be yours.
Commit before you read on
The cluster is idle, the workers are healthy, and total data volume barely moved. Commit before reading on: distinguish the two causes that produce exactly this shape — a straggler from a slow machine, and a straggler from data skew — and name the single measurement that separates them.
Write it down even if you are unsure. An unwritten guess quietly becomes “that is what I thought” the moment you read the answer.