I’ve successfully spun up a BinderHub instance on Google Cloud Platform, and I’m currently trying to make user of a user node pool. I’ve set up a user node pool with a min of 0 and a max of 5 nodes, and I’m trying to now configure things so that these nodes get used. However, I’m a little confused as to where the configuration for the scheduling should go. Currently my
config.yaml file for
helm install jupyterhub/binderhub ... looks like:
config: BinderHub: use_registry: true image_prefix: wwtbinderhub/binder-prod- hub_url: http://... scheduling: userScheduler: enabled: true podPriority: enabled: true userPlaceholder: enabled: true replicas: 3 userPods: nodeAffinity: matchNodePurpose: require cull: enabled: true timeout: 3600 every: 300 singleuser: cpu: limit: 1 guarantee: 0.05 memory: limit: 4G guarantee: 512M
but it wasn’t clear to me whether this is correct, i.e. whether
singleuser should be at the root of the
config.yaml file. Currently if I do the above, the user node pool doesn’t get used, only the default node pool does. Furthermore if I inspect the hub configuration
hub-config through the GCE UI, I don’t see any trace of the custom configuration above. I noticed on this page that hub-specific configuration can go inside:
so I tried putting the sections above (except
config:) inside a
jupyterhub section, i.e.:
binderhub: jupyterhub: scheduling: userScheduler: enabled: true
and re-ran the
help upgrade ... command but no luck.
Could someone clarify where the above sections should go to get picked up?