Repository navigation
Expansion on Misconfigured HPA Fault #136
Description
Currently, the HPA fault is configured to make scale up actions too easy and thus create too many pods for little reason. Although it has SRE applications, this is largely used in FinOps. However, there is an inverse that is currently not explored.
Instead of making scale up actions easy, there should either be an expansion or additional fault with makes aggressive scale down actions (with the idea that this is applied to maybe save cost) which is causing service disruptions because the HPA wants to keep a small number of pods in the face of increased demand.
Activity
[-][SRE][Fault Mechanism]: Expansion on Misconfigured HPA Fault[/-][+]Expansion on Misconfigured HPA Fault[/+]wak327 commentedon Oct 2, 2026
Related to the HPA fault family here: #1255 fixes the existing
misconfigured-kubernetes-horizontal-pod-autoscalerfault, which takes a Deployment/StatefulSet as its argument but was patching the workload rather than its HPA. The fix looks up the HPA throughscaleTargetRef. @hugoev, FYI for #955: that fault takes the HPA directly askubernetesObject, while fault 11 takes the workload. @Red-GV, it might be worth deciding which convention both HPA faults should follow.