Merge pull request #2394 from GoogleCloudPlatform/release-candidate

Release v1.31.0

.github/workflows/pr-precommit.yml

@@ -49,11 +49,5 @@ jobs:
         # https://github.com/terraform-linters/tflint/blob/master/docs/user-guide/plugins.md#avoiding-rate-limiting
         GITHUB_TOKEN: ${{ github.token }}
     - uses: pre-commit/[email protected]
-    - uses: pre-commit-ci/[email protected]
-      # this if statement looks funny but it ensures that this step runs
-      # only if: user has applied "pre-commit-autofix" label
-      # even if: job has failed
-      # not if: job is canceled
-      if: |
-        (success() || failure()) &&
-        contains(github.event.pull_request.labels.*.name, 'pre-commit-autofix')
+      with:
+        extra_args: --show-diff-on-failure --all-files

cmd/create.go

@@ -81,6 +81,7 @@ func runCreateCmd(cmd *cobra.Command, args []string) {
 func doCreate(path string) string {
 	bp := expandOrDie(path)
 	deplDir := filepath.Join(createFlags.outputDir, bp.DeploymentName())
+	logging.Info("Creating deployment folder %q ...", deplDir)
 	checkErr(checkOverwriteAllowed(deplDir, bp, createFlags.overwriteDeployment, createFlags.forceOverwrite))
 	checkErr(modulewriter.WriteDeployment(bp, deplDir))
 	return deplDir

cmd/root.go

@@ -52,7 +52,7 @@ HPC deployments on the Google Cloud Platform.`,
 				logging.Fatal("cmd.Help function failed: %s", err)
 			}
 		},
-		Version:     "v1.30.0",
+		Version:     "v1.31.0",
 		Annotations: annotation,
 	}
 )

community/examples/hpc-slurm-gromacs.yaml

@@ -100,7 +100,7 @@ deployment_groups:
 
   - id: compute_partition
     source: community/modules/compute/schedmd-slurm-gcp-v6-partition
-    use: [compute_nodeset, homefs, appsfs]
+    use: [compute_nodeset]
     settings:
       partition_name: compute
       is_default: true

community/examples/hpc-slurm-ramble-gromacs.yaml

@@ -29,9 +29,9 @@ vars:
 deployment_groups:
 - group: primary
   modules:
-  # Source is an embedded resource, denoted by "resources/*" without ./, ../, /
-  # as a prefix. To refer to a local resource, prefix with ./, ../ or /
-  # Example - ./resources/network/vpc
+  # Source is an embedded module, denoted by "modules/*" without ./, ../, /
+  # as a prefix. To refer to a local module, prefix with ./, ../ or /
+  # Example - ./modules/network/vpc
   - id: network1
     source: modules/network/vpc
 

community/examples/hpc-slurm-sharedvpc.yaml

@@ -65,7 +65,7 @@ deployment_groups:
 
   - id: debug_partition
     source: community/modules/compute/schedmd-slurm-gcp-v6-partition
-    use: [debug_nodeset, homefs]
+    use: [debug_nodeset]
     settings:
       partition_name: debug
       exclusive: false  # allows nodes to stay up after jobs are done
@@ -80,7 +80,7 @@ deployment_groups:
 
   - id: compute_partition
     source: community/modules/compute/schedmd-slurm-gcp-v6-partition
-    use: [compute_nodeset, homefs]
+    use: [compute_nodeset]
     settings:
       partition_name: compute
 

community/examples/hpc-slurm-ubuntu2004-v6.yaml

@@ -31,9 +31,9 @@ vars:
 deployment_groups:
 - group: primary
   modules:
-  # Source is an embedded resource, denoted by "resources/*" without ./, ../, /
-  # as a prefix. To refer to a local resource, prefix with ./, ../ or /
-  # Example - ./resources/network/vpc
+  # Source is an embedded module, denoted by "modules/*" without ./, ../, /
+  # as a prefix. To refer to a local module, prefix with ./, ../ or /
+  # Example - ./modules/network/vpc
   - id: network1
     source: modules/network/vpc
 
@@ -43,7 +43,7 @@ deployment_groups:
     settings:
       local_mount: /home
 
-  - id: debug_node_group
+  - id: debug_nodeset
     source: community/modules/compute/schedmd-slurm-gcp-v6-nodeset
     use: [network1]
     settings:
@@ -54,15 +54,13 @@ deployment_groups:
 
   - id: debug_partition
     source: community/modules/compute/schedmd-slurm-gcp-v6-partition
-    use:
-    - homefs
-    - debug_node_group
+    use: [debug_nodeset]
     settings:
       partition_name: debug
       exclusive: false # allows nodes to stay up after jobs are done
       is_default: true
 
-  - id: compute_node_group
+  - id: compute_nodeset
     source: community/modules/compute/schedmd-slurm-gcp-v6-nodeset
     use: [network1]
     settings:
@@ -72,9 +70,7 @@ deployment_groups:
 
   - id: compute_partition
     source: community/modules/compute/schedmd-slurm-gcp-v6-partition
-    use:
-    - homefs
-    - compute_node_group
+    use: [compute_nodeset]
     settings:
       partition_name: compute
 

community/examples/hpc-slurm-ubuntu2004.yaml

@@ -32,9 +32,9 @@ vars:
 deployment_groups:
 - group: primary
   modules:
-  # Source is an embedded resource, denoted by "resources/*" without ./, ../, /
-  # as a prefix. To refer to a local resource, prefix with ./, ../ or /
-  # Example - ./resources/network/vpc
+  # Source is an embedded module, denoted by "modules/*" without ./, ../, /
+  # as a prefix. To refer to a local module, prefix with ./, ../ or /
+  # Example - ./modules/network/vpc
   - id: network1
     source: modules/network/vpc
 

community/examples/hpc-slurm6-tpu-maxtext.yaml

@@ -70,7 +70,6 @@ deployment_groups:
     source: ./community/modules/compute/schedmd-slurm-gcp-v6-nodeset-tpu
     use: [network]
     settings:
-      name: v4x8
       node_type: v4-8
       tf_version: 2.14.0
       # Preemptible TPUs cost much less than non-preemptible TPUs.

community/examples/hpc-slurm6-tpu.yaml

@@ -32,7 +32,6 @@ deployment_groups:
     source: ./community/modules/compute/schedmd-slurm-gcp-v6-nodeset-tpu
     use: [network]
     settings:
-      name: v3x8
       node_type: v3-8
       tf_version: 2.14.0
       # Preemptible TPUs cost much less than non-preemptible TPUs.

tools/cloud-build/daily-tests/blueprints/slurm-gcp-v5-high-io.yaml → community/examples/htc-slurm-v6.yaml

@@ -1,4 +1,5 @@
-# Copyright 2022 Google LLC
+# Copyright 2024 Google LLC
+# Copyright (C) SchedMD LLC.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -14,22 +15,21 @@
 
 ---
 
-blueprint_name: hpc-cluster-high-io-v5
+# This blueprint provisions a cluster using the Slurm scheduler configured to
+# efficiently run many short duration, loosely-coupled (non-MPI) jobs. See also:
+# https://github.com/GoogleCloudPlatform/slurm-gcp/blob/master/docs/htc.md
+# https://slurm.schedmd.com/high_throughput.html
+
+blueprint_name: htc-slurm-v6
 
 vars:
   project_id:  ## Set GCP Project ID Here ##
-  deployment_name: high-io-slurm-gcp-v5
+  deployment_name: htc-slurm-v6
   region: us-west4
   zone: us-west4-c
   # By default, public IPs are set in the login and controller to allow easier
   # SSH access. To turn this behavior off, set this to true.
   disable_public_ips: false
-  # Set to true for active cluster reconfiguration.
-  # Note that setting this option requires additional dependencies to be installed locally.
-  enable_reconfigure: true
-  # When set, active compute nodes will be cleaned up on destroy.
-  # Note that setting this option requires additional dependencies to be installed locally.
-  enable_cleanup_compute: true
 
 # Documentation for each of the modules used below can be found at
 # https://github.com/GoogleCloudPlatform/hpc-toolkit/blob/main/modules/README.md
@@ -40,18 +40,18 @@ deployment_groups:
   # Source is an embedded module, denoted by "modules/*" without ./, ../, /
   # as a prefix. To refer to a local or community module, prefix with ./, ../ or /
   # Example - ./modules/network/pre-existing-vpc
-  - id: network1
+  - id: network
     source: modules/network/vpc
 
   - id: homefs
     source: modules/file-system/filestore
-    use: [network1]
+    use: [network]
     settings:
       local_mount: /home
 
   - id: projectsfs
     source: modules/file-system/filestore
-    use: [network1]
+    use: [network]
     settings:
       filestore_tier: HIGH_SCALE_SSD
       size_gb: 10240
@@ -61,94 +61,98 @@ deployment_groups:
   # https://console.developers.google.com/marketplace/product/ddnstorage/exascaler-cloud
   - id: scratchfs
     source: community/modules/file-system/DDN-EXAScaler
-    use: [network1]
+    use: [network]
     settings:
       local_mount: /scratch
 
+  # The compute partition is designed for performance.
+  # Use:
+  # `srun -N 4 -p compute <<Command>>` for any node in the partition.
+  # `srun -N 4 -p compute --mincpus 30 <<Command>>` for node group c2s60.
+  - id: compute_nodeset_c2s60
+    source: community/modules/compute/schedmd-slurm-gcp-v6-nodeset
+    use: [network]
+    settings:
+      name: c2s60
+      node_count_dynamic_max: 200
+      bandwidth_tier: gvnic_enabled
+      enable_placement: false
+
+  - id: compute_nodeset_c2s30
+    source: community/modules/compute/schedmd-slurm-gcp-v6-nodeset
+    use: [network]
+    settings:
+      node_count_dynamic_max: 200
+      machine_type: c2-standard-30
+      bandwidth_tier: gvnic_enabled
+      enable_placement: false
+
+  - id: compute_partition
+    source: community/modules/compute/schedmd-slurm-gcp-v6-partition
+    use:
+    - compute_nodeset_c2s60
+    - compute_nodeset_c2s30
+    settings:
+      partition_name: compute
+      exclusive: false
+
   # The lowcost partition is designed to run at a lower cost and without additional quota
   # Use:
   # `srun -N 4 <<Command>>` for any node in the partition.
   # `srun -N 4 --mincpus 2` for node group n2s4.
-  - id: low_cost_node_group_n2s2
-    source: community/modules/compute/schedmd-slurm-gcp-v5-node-group
+  - id: low_cost_nodeset_n2s2
+    source: community/modules/compute/schedmd-slurm-gcp-v6-nodeset
+    use: [network]
     settings:
       name: n2s2
       machine_type: n2-standard-2
       node_count_dynamic_max: 10
+      bandwidth_tier: gvnic_enabled
+      enable_placement: false
 
-  - id: low_cost_node_group_n2s4
-    source: community/modules/compute/schedmd-slurm-gcp-v5-node-group
+  - id: low_cost_nodeset_n2s4
+    source: community/modules/compute/schedmd-slurm-gcp-v6-nodeset
+    use: [network]
     settings:
       name: n2s4
       machine_type: n2-standard-4
       node_count_dynamic_max: 10
+      bandwidth_tier: gvnic_enabled
+      enable_placement: false
 
   - id: low_cost_partition
-    source: community/modules/compute/schedmd-slurm-gcp-v5-partition
+    source: community/modules/compute/schedmd-slurm-gcp-v6-partition
     use:
-    - network1
-    - homefs
-    - scratchfs
-    - projectsfs
-    - low_cost_node_group_n2s2
-    - low_cost_node_group_n2s4
+    - low_cost_nodeset_n2s2
+    - low_cost_nodeset_n2s4
     settings:
       is_default: true
       partition_name: lowcost
-      enable_placement: false
       exclusive: false
 
-  # The compute partition is designed for performance.
-  # Use:
-  # `srun -N 4 -p compute <<Command>>` for any node in the partition.
-  # `srun -N 4 -p compute --mincpus 30 <<Command>>` for node group c2s60.
-
-  - id: compute_node_group_c2s60
-    source: community/modules/compute/schedmd-slurm-gcp-v5-node-group
-    settings:
-      name: c2s60
-      node_count_dynamic_max: 200
-
-  - id: compute_node_group_c2s30
-    source: community/modules/compute/schedmd-slurm-gcp-v5-node-group
-    settings:
-      name: c2s30
-      node_count_dynamic_max: 200
-      machine_type: c2-standard-30
-
-  - id: compute_partition
-    source: community/modules/compute/schedmd-slurm-gcp-v5-partition
-    use:
-    - network1
-    - homefs
-    - scratchfs
-    - projectsfs
-    - compute_node_group_c2s60
-    - compute_node_group_c2s30
+  - id: slurm_login
+    source: community/modules/scheduler/schedmd-slurm-gcp-v6-login
+    use: [network]
     settings:
-      partition_name: compute
+      name_prefix: login
+      machine_type: n2-standard-4
+      disable_login_public_ips: $(vars.disable_public_ips)
 
   - id: slurm_controller
-    source: community/modules/scheduler/schedmd-slurm-gcp-v5-controller
+    source: community/modules/scheduler/schedmd-slurm-gcp-v6-controller
     use:
-    - network1
+    - network
     - homefs
     - scratchfs
     - projectsfs
     - low_cost_partition
     - compute_partition
+    - slurm_login
     settings:
       machine_type: c2-standard-8
       disable_controller_public_ips: $(vars.disable_public_ips)
-
-  - id: slurm_login
-    source: community/modules/scheduler/schedmd-slurm-gcp-v5-login
-    use:
-    - network1
-    - slurm_controller
-    settings:
-      machine_type: n2-standard-4
-      disable_login_public_ips: $(vars.disable_public_ips)
+      slurm_conf_tpl: modules/embedded/community/modules/scheduler/schedmd-slurm-gcp-v6-controller/etc/htc-slurm.conf.tpl
+      slurmdbd_conf_tpl: modules/embedded/community/modules/scheduler/schedmd-slurm-gcp-v6-controller/etc/htc-slurmdbd.conf.tpl
 
   - id: hpc_dashboard
     source: modules/monitoring/dashboard

community/examples/intel/hpc-slurm-daos.yaml

@@ -131,7 +131,7 @@ deployment_groups:
 
   - id: debug_partition
     source: community/modules/compute/schedmd-slurm-gcp-v6-partition
-    use: [debug_nodeset, homefs]
+    use: [debug_nodeset]
     settings:
       partition_name: debug
       exclusive: false  # allows nodes to stay up after jobs are done
@@ -154,7 +154,7 @@ deployment_groups:
 
   - id: compute_partition
     source: community/modules/compute/schedmd-slurm-gcp-v6-partition
-    use: [compute_nodeset, homefs]
+    use: [compute_nodeset]
     settings:
       partition_name: compute
 

community/examples/tutorial-starccm-slurm.yaml

@@ -28,9 +28,9 @@ vars:
 deployment_groups:
 - group: primary
   modules:
-  # Source is an embedded resource, denoted by "resources/*" without ./, ../, /
-  # as a prefix. To refer to a local resource, prefix with ./, ../ or /
-  # Example - ./resources/network/vpc
+  # Source is an embedded module, denoted by "modules/*" without ./, ../, /
+  # as a prefix. To refer to a local module, prefix with ./, ../ or /
+  # Example - ./modules/network/vpc
   - id: network1
     source: modules/network/vpc