DevOps Dictionary

Karpenter

Karpenter is an open source Kubernetes node autoscaler that automatically provisions and retires worker nodes so cluster capacity matches workload demand. It monitors the Kubernetes scheduler for pods (deployable workload units) that are pending because no existing node has the right CPU, memory, or other requirements, then requests new instances that fit those pods and later consolidates by draining and terminating underused nodes. This gap exists because Kubernetes decides where pods should run, but it does not create or resize the underlying machines on its own. With Karpenter, teams typically get faster scale-out and better packing of workloads onto fewer, right-sized nodes; without it, they often rely on slower, rule-based node groups or manual changes, which can leave pods stuck pending during spikes or keep excess nodes running when demand drops.

A
B
C
D
E
F
G
H
I
J
K
L
M
N
O
P
Q
R
S
T
U
V
W
Y
X
Z