nodus cloud connect

Run your work in your own cloud account with its API key (Beta)

Synopsis

Connect your own cloud account so Nodus places your work there first, within a credit limit, and uses Nodus capacity when it runs out (see –burst). CLOUD is lambda, runpod, digitalocean, thunder, nebius, denvr, jarvis or massedcompute, or aws, gcp or azure, which take an account already connected whose execute stack or script has run and read no key; on GCP, –gcp-project picks one of its projects (default the first). kubernetes runs each machine as a pod in one namespace of your own GPU cluster: apply the Nodus manifest there first, then pass its ServiceAccount’s kubeconfig with –kubeconfig-file, the namespace, and each GPU type with your own hourly rate per GPU as –gpu TYPE=USD[,LABEL=VALUE…], the node labels that pick its nodes; only the kubeconfig’s current server, certificate authority and token are sent. –scale-up-wait sets how long a pod no node has room for waits for your autoscaler (default 10 minutes, 0 for a fixed-size pool). The API key (the service account’s private key for nebius) is read from standard input, or from the environment variable –key-env names; it is never printed and Nodus never returns it. –field gives the key’s other fields, such as projectID and region for nebius, cluster for denvr or scriptID for jarvis.

nodus cloud connect CLOUD NAME [flags]

Examples

  nodus cloud connect lambda my-lambda --credit-limit 500 --ssh-key my-key --key-env LAMBDA_API_KEY
  nodus cloud connect aws aws-main --credit-limit 500
  nodus cloud connect gcp gcp-main --credit-limit 1000 --gcp-project my-training-project
  nodus cloud connect azure azure-main --credit-limit 1000
  nodus cloud connect kubernetes gpu-cluster --kubeconfig-file nodus.kubeconfig --namespace nodus \
    --gpu H100-SXM=2.10 --credit-limit 2000
  nodus cloud connect denvr my-denvr --credit-limit 300 --field cluster=Msc1 --key-env DENVR_API_KEY
  printf '%s' "$RUNPOD_API_KEY" | nodus cloud connect runpod my-runpod --credit-limit 200

Options

      --burst string             When the account is full or out of credit: Allow (default), Approve or Deny Nodus capacity
      --credit-limit string      Most cloud spend in USD Nodus may cause in the account per period
      --field stringToString     Another field of the key, as name=value (repeatable) (default [])
      --gcp-project string       GCP project to run work in, one of the account's (default the first)
      --gpu stringArray          Kubernetes: a GPU type and your hourly USD per GPU, with the node labels that pick its nodes: TYPE=USD[,LABEL=VALUE...]; repeat per type (required on create)
  -h, --help                     help for connect
      --key-env string           Environment variable holding the API key, instead of standard input
      --kubeconfig-file string   Kubernetes: the Nodus ServiceAccount's kubeconfig file, or - for standard input
      --max-hourly string        Most USD per hour of machines running there at once (default the credit limit)
      --namespace string         Kubernetes: the namespace Nodus runs pods in, where its manifest is applied
      --period string            Monthly (default) or Total
      --priority int32           Order among your accounts: 1 takes work first; unset comes after every numbered account
      --region string            Kubernetes: the country the cluster runs in, such as us or fi (default us)
      --scale-up-wait int        Kubernetes: minutes a pod no node has room for waits for your autoscaler, 0 to 20 (default 10; 0 for a fixed-size pool) (default -1)
      --ssh-key string           SSH key the cloud installs on machines (Lambda: key name; DigitalOcean: fingerprint)

Options inherited from parent commands

      --context string   Context from the config file to use
      --org string       Organization (selects the context for that org)
  -p, --project string   Project to work in
  -v, --verbose          Log each API request (never credentials)

nodus cloud connect