Archive Topic Map
Scheduling & Placement
Scheduling is economics and ethics: who runs where, at what cost, under what constraints, and what happens when scarcity arrives.
Orientation
A curated shelf for study and for retrieval.
Requests are promises
The scheduler uses requests to decide placement. If your requests are dishonest, you get a dishonest cluster: noisy neighbors, eviction cascades, and capacity plans built on sand.
Placement constraints are powerful. They are also easy to make impossible. Learn the difference between ‘preferred’ and ‘required’, and treat taints as governance.
- Requests drive placement; limits shape runtime behavior and throttling.
- Affinity and topology constraints can silently eliminate all nodes.
- Priority and preemption are governance tools; misuse causes conflict and instability.
Core texts
Scheduler realities and the discipline of scarcity.
Tenet
TenetTenet III: Scheduling
Placement, efficiency, fairness, and resource stewardship.
Text
Codex GigasCapacity, Bin Packing, and the Lies We Tell the Scheduler
How clusters become inefficient—and how to recover with discipline.
Text
Codex GigasThe Scheduler Under Scarcity
Priority, preemption, fairness, and when scarcity becomes governance.
Practice and diagnostics
Read scheduler events; fix constraints and economics.
Related maps
Adjacent shelves for continued study.
Topic map
MapWorkloads
Continue with the adjacent shelf.
Topic map
MapOperations
Continue with the adjacent shelf.
Topic map
MapTroubleshooting
Continue with the adjacent shelf.
Canonical link
Canonical URL: /library/topics/scheduling