NodeSelector for Jobs dispatched in Kubernetes

Hey, Im facing this weird issue that when a build is triggered, the pod is created in a random node and when the cpu usage becomes high, for example by composer install, the node in which the job started, crashes.
Is it possible to add resource limits to the job pod? or Atleast NodeSelector so I can mention which node pool is reserved only for DroneCI?

2 Likes

Node selection can be configured in the yaml, and maps to the Kubernetes nodeSelector:
https://docs.drone.io/user-guide/pipeline/nodes/

Resource limits can also be configured in the yaml:

---
kind: pipeline
name: default

platform:
  os: linux
  arch: amd64

steps:
- name: build
  image: golang
  commands:
  - go build
  resources:
    limits:
      cpu: 2000
      memory: 100MiB
    requests:
      cpu: 100
      memory: 50MiB

NOTE: this will only work with the very latest rc.4 snapshot.

2 Likes