Cluster Configuration

Minimal configuration

The simplest valid configuration creates a cluster with one controller and one worker using the default image:

kind: Cluster

This is equivalent to the fully expanded form:

kind: Cluster
name: default
defaults:
  image: ghcr.io/gsi-hpc/sind-node:latest
  cpus: 1
  memory: 512m
  tmpSize: 256m
nodes:
  - role: controller
  - role: worker

Full example

kind: Cluster
name: test-cluster

defaults:
  image: ghcr.io/gsi-hpc/sind-node:latest
  cpus: 1
  memory: 512m
  tmpSize: 256m

storage:
  dataStorage:
    type: volume
    mountPath: /data

slurm:
  main: |
    SelectType=select/cons_tres
    SelectTypeParameters=CR_Core_Memory
  cgroup: |
    ConstrainCores=yes

nodes:
  - role: controller
    cpus: 2
    memory: 1g
    tmpSize: 512m

  - role: submitter

  - role: worker
    count: 3
    cpus: 2
    memory: 1g

  - role: worker
    count: 2
    managed: false

Top-level fields

Field	Required	Default	Description
`kind`	yes	—	Must be `"Cluster"`
`name`	no	`"default"`	Cluster name, used in resource naming
`realm`	no	`"sind"`	Realm namespace for resource isolation
`defaults`	no	—	Default settings applied to all nodes
`storage`	no	—	Shared storage configuration
`slurm`	no	—	Slurm configuration extension
`nodes`	no	1 controller + 1 worker	Node definitions

Defaults section

The defaults section sets values inherited by all nodes unless overridden at the node level.

Field	Default	Description
`image`	`ghcr.io/gsi-hpc/sind-node:latest`	Container image
`cpus`	`1`	CPU limit per container
`memory`	`"512m"`	Memory limit per container
`tmpSize`	`"256m"`	tmpfs size for `/tmp`
`capAdd`	none	Extra Linux capabilities
`capDrop`	none	Dropped Linux capabilities
`devices`	none	Host devices to expose
`securityOpt`	none	Extra security options

Scalar fields (image, cpus, memory, tmpSize) are overridden by per-node values. List fields (capAdd, capDrop, devices, securityOpt) are merged with per-node values.

Storage section

storage:
  dataStorage:
    type: volume       # "volume" or "hostPath"
    hostPath: ./data   # required if type is hostPath
    mountPath: /data   # default: /data

Field	Default	Description
`type`	`"volume"`	`"volume"` for a Docker volume, `"hostPath"` for a bind mount
`hostPath`	—	Path on the host (required when `type: hostPath`)
`mountPath`	`"/data"`	Mount point inside containers

Slurm section

The slurm section extends the generated Slurm configuration. Each key maps to a config file:

Key	Config file	sind generates defaults
`main`	`slurm.conf`	yes
`cgroup`	`cgroup.conf`	yes
`gres`	`gres.conf`	no
`topology`	`topology.conf`	no
`plugstack`	`plugstack.conf`	yes (always scaffolded)

Each key supports two forms:

String form — content appended to the config file:

slurm:
  main: |
    SelectType=select/cons_tres
    SelectTypeParameters=CR_Core_Memory
  cgroup: |
    ConstrainCores=yes

Map form — named fragments placed in a .conf.d/ directory:

slurm:
  main:
    scheduling: |
      SchedulerType=sched/backfill
      SchedulerParameters=bf_continue
    resources: |
      SelectType=select/cons_tres

Fragment validation:

Names must be plain filenames (no path separators)
Names and content must not be empty

See Slurm Configuration for details on the generated files.

Validation rules

kind must be "Cluster"
Exactly one controller node is required
At most one submitter node is allowed
At least one worker node is required
count is only valid for worker nodes
managed is only valid for worker nodes
count must not be negative
capAdd/capDrop values must be recognized Linux capability names (e.g. SYS_ADMIN, NET_ADMIN, ALL)
devices paths must be absolute (start with /)