Spot‑Enabled Node Pool Toleration Injection for Low‑Priority Workloads in Rancher
0 reputation · 22 Dec 2020, 13:00 UTC
Spot‑Enabled Node Pool Toleration Injection for Low‑Priority Workloads in Rancher
The goal is to allow low‑traffic workloads to run on spot instances to reduce cost while preserving availability.
When a node pool uses a spot‑enabled template, Rancher does not automatically add the `spot` toleration to pods, leaving the decision to the pod specification or to administrators who must manually edit workloads.
It is unclear whether Rancher should provide a cluster‑wide option to inject the toleration for workloads marked low‑priority, or whether manual toleration management should remain the preferred approach to keep cost‑versus‑availability control explicit.
- Should Rancher offer a setting that automatically injects the `spot` toleration for low‑priority workloads when a spot‑enabled node pool exists?
- What RBAC permissions are required for administrators to enable, disable, or override this automatic injection?
- How should Rancher behave when a workload already defines tolerations to avoid conflicts or duplicate entries?