Knowledgebase

Handling Pod Scheduling and Placement Print

  • 0

Deciding where workloads run.

WHAT THE SCHEDULER USES

Resource requests Node capacity Constraints you specify

WHAT A NODE SELECTOR DOES

Restricts a pod to nodes with a label.

WHERE THAT IS USED

Nodes with particular hardware, or in a particular zone.

WHAT AFFINITY ADDS

Preferences as well as requirements.

WHY PREFERENCES MATTER

A hard requirement that cannot be met leaves the pod pending forever.

WHAT ANTI-AFFINITY DOES

Keeps pods apart.

WHY THAT MATTERS

Several replicas on one node all die when it does.

WHAT TO SET FOR ANYTHING IMPORTANT

Anti-affinity spreading replicas across nodes.

WHAT TAINTS AND TOLERATIONS DO

Reserve nodes for specific workloads.

HOW

A taint repels pods; a toleration permits one to land anyway.

WHERE THAT SUITS

Nodes with expensive hardware Nodes reserved for system components

WHAT A DISRUPTION BUDGET DOES

Limits how many replicas may be unavailable during voluntary operations.

WHY IT MATTERS

Node maintenance otherwise removes every replica at once.

WHAT TO SET

A minimum available, for anything serving traffic.

WHAT TO BE CAREFUL WITH

Budgets so strict that node maintenance cannot proceed.

WHAT TO CHECK WHEN A POD WILL NOT SCHEDULE

Its events, which state the reason.


Was this answer helpful?
Back

Are you happy with your experience? Leave us a review on Trustpilot.


Trustpilot