Kubernetes executor: node tolerations per job as environment variables
Looking for a way of running some jobs on a nodes with specific taints, thought it may be possible to override the per job taints like we can do with annotations and limits/requests (KUBERNETES_POD_ANNOTATIONS_1
or KUBERNETES_MEMORY_LIMIT
).
Tried something similar to pod_annotations, but it's not working:
variables:
KUBERNETES_NODE_TOLERATIONS_1: "LargeInstance=1:NoSchedule"
Any plans adding support for node_tolerations
the same way as pod_annotations
?