feat(helm): add initial dask support (#821)

reanahub · Oct 24, 2024 · 7729cc7 · 7729cc7
1 parent 467be7a
commit 7729cc7
Show file tree

Hide file tree

Showing 6 changed files with 54 additions and 0 deletions.
diff --git a/helm/reana/Chart.yaml b/helm/reana/Chart.yaml
@@ -43,3 +43,9 @@ dependencies:
     version: 0.47.7
     repository: https://fluent.github.io/helm-charts
     condition: fluent-bit.enabled
+  - name: dask-kubernetes-operator
+    version: 2024.8.0
+    repository: https://helm.dask.org
+    condition: dask.enabled
+    tags:
+      - dask
diff --git a/helm/reana/README.md b/helm/reana/README.md
@@ -68,6 +68,11 @@ This Helm automatically prefixes all names using the release name to avoid colli
 | `components.reana_workflow_engine_snakemake.environment` | [REANA-Workflow-Engine-Snakemake](https://github.com/reanahub/reana-workflow-engine-snakemake) environment variables | `{}`            |
 | `components.reana_workflow_engine_snakemake.image`       | [REANA-Workflow-Engine-Snakemake image](https://hub.docker.com/r/reanahub/reana-workflow-engine-snakemake) to use | `docker.io/reanahub/reana-workflow-engine-snakemake:<chart-release-version>` |
 | `compute_backends`                                       | List of supported compute backends (kubernetes, htcondorcern, slurmcern)             | "kubernetes"                                    |
+| `dask.enabled` | Install dask-kubernetes-operator custom resources in the cluster to support Dask workflows | false |
+| `dask.cluster_max_memory_limit` | Max memory limit for Dask clusters | "16Gi" | 
+| `dask.cluster_default_number_of_workers` | Number of workers in Dask clusters by default | 2 | 
+| `dask.cluster_default_single_worker_memory` | Amount of memory used by a single Dask worker by default | "2Gi" |
+| `dask.cluster_max_single_worker_memory` | Maximum amount of memory that can be used by a single Dask worker | "8Gi" |
 | `db_env_config.REANA_DB_HOST`                            | Environment variable to connect to external databases                                | `<chart-release-name>-db`                       |
 | `db_env_config.REANA_DB_NAME`                            | Environment variable to connect to external databases                                | reana                                           |
 | `db_env_config.REANA_DB_PORT`                            | Environment variable to connect to external databases                                | "5432"                                          |

diff --git a/helm/reana/templates/reana-server.yaml b/helm/reana/templates/reana-server.yaml
@@ -98,6 +98,18 @@ spec:
             value: !!str {{ .Values.kubernetes_jobs_timeout_limit | default 604800 }}  # 604800 seconds = 7 days
           - name: REANA_KUBERNETES_JOBS_MEMORY_LIMIT
             value: {{ .Values.kubernetes_jobs_memory_limit | default "4Gi" }}
+          - name: DASK_ENABLED
+            value: {{ .Values.dask.enabled | quote }}
+          {{- if .Values.dask.enabled }}
+          - name: REANA_DASK_CLUSTER_MAX_MEMORY_LIMIT
+            value: {{ .Values.dask.cluster_max_memory_limit | default "16Gi" }}
+          - name: REANA_DASK_CLUSTER_DEFAULT_NUMBER_OF_WORKERS
+            value: !!str {{ .Values.dask.cluster_default_number_of_workers | default 2 }}
+          - name: REANA_DASK_CLUSTER_DEFAULT_SINGLE_WORKER_MEMORY
+            value: {{ .Values.dask.cluster_default_single_worker_memory | default "2Gi" }}
+          - name: REANA_DASK_CLUSTER_MAX_SINGLE_WORKER_MEMORY
+            value: {{ .Values.dask.cluster_max_single_worker_memory | default "8Gi" }}
+          {{- end }}          
           - name: WORKSPACE_PATHS
             value: {{ .Values.workspaces.paths | toJson | quote }}
           - name: WORKSPACE_RETENTION_PERIOD

diff --git a/helm/reana/templates/reana-workflow-controller.yaml b/helm/reana/templates/reana-workflow-controller.yaml
@@ -179,6 +179,18 @@ spec:
             value: {{ .Values.components.reana_job_controller.environment | toJson | quote }}
           - name: REANA_INTERACTIVE_SESSIONS_ENVIRONMENTS
             value: {{ .Values.interactive_sessions.environments | toJson | quote }}
+          - name: DASK_ENABLED
+            value: {{ .Values.dask.enabled | quote }}
+          {{- if .Values.dask.enabled }}
+          - name: REANA_DASK_CLUSTER_MAX_MEMORY_LIMIT
+            value: {{ .Values.dask.cluster_max_memory_limit | default "16Gi" }}
+          - name: REANA_DASK_CLUSTER_DEFAULT_NUMBER_OF_WORKERS
+            value: !!str {{ .Values.dask.cluster_default_number_of_workers | default 2 }}
+          - name: REANA_DASK_CLUSTER_DEFAULT_SINGLE_WORKER_MEMORY
+            value: {{ .Values.dask.cluster_default_single_worker_memory | default "2Gi" }}
+          - name: REANA_DASK_CLUSTER_MAX_SINGLE_WORKER_MEMORY
+            value: {{ .Values.dask.cluster_max_single_worker_memory | default "8Gi" }}
+          {{- end }}        
           {{- if .Values.reana_hostname }}
           - name: REANA_HOSTNAME
             value: {{ .Values.reana_hostname }}

diff --git a/helm/reana/templates/roles.yaml b/helm/reana/templates/roles.yaml
@@ -22,6 +22,18 @@ rules:
 - apiGroups: ["metrics.k8s.io"]
   resources: ["pods", "nodes"]
   verbs: ["get", "list", "watch"]
+# Custom dask kubernetes resources
+{{- if .Values.dask.enabled }}
+- apiGroups: ["kubernetes.dask.org"]
+  resources: ["daskclusters", "daskautoscalers"]
+  verbs: ["create", "get", "list", "watch", "update", "patch", "delete"]
+{{- end }}
+# Traefik Middleware permissions
+{{- if and .Values.traefik.enabled .Values.dask.enabled }}
+- apiGroups: ["traefik.io"]
+  resources: ["middlewares"]
+  verbs: ["get", "list", "watch", "create", "update", "patch", "delete"]
+{{- end }}
 ---
 apiVersion: rbac.authorization.k8s.io/v1
 kind: ClusterRoleBinding

diff --git a/helm/reana/values.yaml b/helm/reana/values.yaml
@@ -178,6 +178,13 @@ traefik:
     enabled: true
     isDefaultClass: true
 
+dask:
+  enabled: true
+  cluster_max_memory_limit: "16Gi"
+  cluster_default_number_of_workers: 2
+  cluster_default_single_worker_memory: "2Gi"
+  cluster_max_single_worker_memory: "4Gi"
+
 pgbouncer:
   enabled: false
   image: docker.io/bitnami/pgbouncer:1.23.1