30 feature update slurm version to 22052 (#33)

* Update slurm from 21.08.8 to 22.05.2 Fix up some issues found during testing. Use the correct Rocky Linux AMIs. Fix an selinux issue with Alma Linux. * Add extra 5 GB to root volume of compute node AMIs so don't fill up AMI creation failed because the root volume filled up. The Alma Linux 8 arm64 AMI only had a 4 GB root volume and it filled up when installing packages.
aws-samples · Jun 22, 2022 · 57da982 · 57da982
1 parent 48fa695
commit 57da982
Show file tree

Hide file tree

Showing 8 changed files with 197 additions and 157 deletions.
diff --git a/.gitignore b/.gitignore
@@ -1,4 +1,8 @@
 
+.mkdocs_venv/
+site/
+.vscode/
+
 # Jekyll
 Gemfile.lock
 .jekyll-cache

diff --git a/install.sh b/install.sh
@@ -12,17 +12,24 @@ deactivate &> /dev/null || true
 
 if ! yum list installed make &> /dev/null; then
     echo -e "\nInstalling make"
-    sudo yum -y install make
+    if ! sudo yum -y install make; then
+        echo -e "\nwarning: Couldn't install make"
+    fi
 fi
 
 if ! yum list installed wget &> /dev/null; then
     echo -e "\nInstalling wget"
-    sudo yum -y install wget
+    if ! sudo yum -y install wget; then
+        echo -e "\nwarning: Couldn't install wget"
+    fi
 fi
 
 if ! python3 --version &> /dev/null; then
     echo -e "\nInstalling python3"
-    sudo yum -y install python3
+    if ! sudo yum -y install python3; then
+        echo -e "\nerror: Couldn't find python3 in the path or install it. This is required."
+        exit 1
+    fi
 fi
 
 # Check python version
@@ -71,8 +78,8 @@ if ! cdk --version &> /dev/null; then
 fi
 version=$(cdk --version | awk '{print $1}')
 if [[ $version != $CDK_VERSION ]]; then
-    echo "Updating the local version of aws-cdk from version $version to $CDK_VERSION"
-    sudo npm update -g aws-cdk@$CDK_VERSION
+    echo "Updating the global version of aws-cdk from version $version to $CDK_VERSION"
+    sudo npm install -g aws-cdk@$CDK_VERSION
 fi
 
 # Create python virtual environment

diff --git a/source/cdk/cdk_slurm_stack.py b/source/cdk/cdk_slurm_stack.py
@@ -2188,10 +2188,13 @@ def create_slurm_node_ami(self):
                             logger.error(f"AmiMap doesn't have ImageId for {self.region}/{distribution}/{distribution_major_version}/{architecture}")
                             exit(1)
                     ami_info = ec2_client.describe_images(ImageIds=[ami_id])['Images'][0]
+                    root_device_name = ami_info['RootDeviceName']
                     block_devices = []
-                    root_device = True
                     for block_device_info in ami_info['BlockDeviceMappings']:
-                        ami_volume_size = int(block_device_info['Ebs']['VolumeSize'])
+                        device_name = block_device_info['DeviceName']
+                        root_device = root_device_name == device_name
+                        # Add 5GB to the AMI's root device so that there is room for installing packages
+                        ami_volume_size = int(block_device_info['Ebs']['VolumeSize']) + 5
                         if root_device:
                             try:
                                 volume_size = str(self.config['slurm']['SlurmNodeAmis']['BaseAmis'][self.region][distribution][distribution_major_version][architecture]['RootDeviceSize'])
@@ -2204,7 +2207,6 @@ def create_slurm_node_ami(self):
                                         exit(1)
                             except KeyError:
                                 volume_size = ami_volume_size
-                            root_device = False
                         else:
                             volume_size = block_device_info['Ebs']['VolumeSize']
                         block_devices.append(

diff --git a/source/cdk/config_schema.py b/source/cdk/config_schema.py
@@ -23,6 +23,8 @@
 from schema import Schema, And, Use, Optional, Regex, SchemaError
 from sys import exit
 
+DEFAULT_SLURM_VERSION = '22.05.2'
+
 config = {}
 
 # Determine all AWS regions available on the account. We do not display opt-out region
@@ -127,7 +129,7 @@
             # SlurmVersion:
             #     Latest tested version
             #     Critical security fix released in 21.08.8. Must be later than that.
-            Optional('SlurmVersion', default='21.08.8'): str,
+            Optional('SlurmVersion', default=DEFAULT_SLURM_VERSION): str,
             #
             # ClusterName:
             #     Default to the StackName

diff --git a/source/create-ami-map.py b/source/create-ami-map.py
@@ -23,6 +23,7 @@
 '''
 import argparse
 import boto3
+import json
 import logging
 from pprint import PrettyPrinter
 
@@ -32,11 +33,13 @@
 
 distributions_dict = {
     'AlmaLinux': {
-        # x86_64: https://aws.amazon.com/marketplace/pp/prodview-mku4y3g4sjrye
-        # arm_64: https://aws.amazon.com/marketplace/pp/prodview-zgsymdwitnxmm
         'owner': '679593333241',
         'major_versions': ['8'],
         'name_filter': 'AlmaLinux OS {distribution_major_version}*',
+        'product_codes': [
+            'be714bpjscoj5uvqz0of5mscl', # x86_64: https://aws.amazon.com/marketplace/pp/prodview-mku4y3g4sjrye
+            '6vto7uou7jjh4um1mwoy8ov0s', # arm_64: https://aws.amazon.com/marketplace/pp/prodview-zgsymdwitnxmm
+        ]
     },
     'Amazon': {
         'owner': '137112412989',
@@ -46,19 +49,20 @@
     'CentOS': {
         'owner': '125523088429',
         'major_versions': ['7', '8'],
-        'name_filter': 'CentOS {distribution_major_version}*'
+        'name_filter': 'CentOS {distribution_major_version}*',
     },
     'RedHat': {
         'owner': '309956199498',
         'major_versions': ['7', '8'],
         'name_filter': 'RHEL-{distribution_major_version}*'
     },
     'Rocky': {
-        # x86_64: https://aws.amazon.com/marketplace/pp/prodview-2otariyxb3mqu
-        # arm_64: https://aws.amazon.com/marketplace/pp/prodview-uzg6o44ep3ugw
         'owner': '679593333241',
         'major_versions': ['8'],
-        'name_filter': 'Rocky Linux {distribution_major_version}*'
+        'product_codes': [
+            'cotnnspjrsi38lfn8qo4ibnnm', # x86_64: https://aws.amazon.com/marketplace/pp/prodview-2otariyxb3mqu
+            '7tvwi95pv43herd5jg0bs6cu5', # arm_64: https://aws.amazon.com/marketplace/pp/prodview-uzg6o44ep3ugw
+        ]
     },
 }
 
@@ -90,17 +94,30 @@ def main(filename, region, distribution):
             for distribution_major_version in distributions_dict[distribution]['major_versions']:
                 logger.debug(f"distribution_major_version: {distribution_major_version}")
                 ami_map[region][distribution][distribution_major_version] = {}
-                name_filter = distributions_dict[distribution]['name_filter'].format(distribution_major_version=distribution_major_version)
-                logger.debug(f"name_filter: {name_filter}")
                 kwargs = {
                     'Owners': [distributions_dict[distribution]['owner']],
                     'Filters': [
-                        {'Name': 'name', 'Values': [name_filter]},
                         {'Name': 'state', 'Values': ['available']}
                     ]
                 }
+                if 'name_filter' in distributions_dict[distribution]:
+                    name_filter = distributions_dict[distribution]['name_filter'].format(distribution_major_version=distribution_major_version)
+                    logger.debug(f"name_filter: {name_filter}")
+                    filter = {
+                        'Name': 'name',
+                        'Values': [name_filter]
+                    }
+                    kwargs['Filters'].append(filter)
+                if 'product_codes' in distributions_dict[distribution]:
+                    product_codes = distributions_dict[distribution]['product_codes']
+                    logger.debug(f'product_codes: {product_codes}')
+                    filter = {
+                        'Name': 'product-code',
+                        'Values': product_codes
+                    }
+                    kwargs['Filters'].append(filter)
                 images = ec2_client.describe_images(**kwargs).get('Images', None)
-                logger.debug(f"Found {len(images)} images")
+                logger.debug(f"Found {len(images)} images:\n{json.dumps(images, indent=4)}")
                 for image in images:
                     if 'BETA' in image['Name']:
                         continue
@@ -117,7 +134,7 @@ def main(filename, region, distribution):
         if not ami_map[region]:
             del ami_map[region]
 
-    logger.debug(pp.pformat(ami_map))
+    logger.debug(f"ami_map:\n{json.dumps(ami_map, indent=4)}")
 
     fh = open(filename, 'w')
     print("AmiMap:", file=fh)