[Develop] Support for RHEL9 and Rocky9

.github/workflows/dokken-system-tests.yml

@@ -35,6 +35,8 @@ jobs:
           - ubuntu2204
           - rhel8
           - rocky8
+          - rhel9
+          - rocky9
       fail-fast: false
     steps:
       - uses: actions/checkout@main

CHANGELOG.md

@@ -7,6 +7,8 @@ This file is used to list changes made in each version of the AWS ParallelCluste
 ------
 
 **ENHANCEMENTS**
+- Add support for RHEL9.
+- Add support for Rocky Linux 9 as `CustomAmi` created through `build-image` process. No public official ParallelCluster Rocky9 Linux AMI is made available at this time.
 - Add the configuration parameter `DeploymentSettings/DefaultUserHome` to allow users to move the default user's home directory to `/local/home` instead of `/home` (default).
 - Add possibility to choose between Open and Closed Source Nvidia Drivers when building an AMI, through the ```['cluster']['nvidia']['kernel_open']``` cookbook node attribute.
 

cookbooks/aws-parallelcluster-environment/files/default/ec2_udev_rules/ec2_dev_2_volid.py

@@ -1,31 +1,11 @@
 # FIXME: Fix Code Duplication
 # pylint: disable=R0801
 
-import configparser
+import json
 import os
 import re
 import sys
 import syslog
-import time
-
-import boto3
-import requests
-from botocore.config import Config
-
-METADATA_REQUEST_TIMEOUT = 60
-
-
-def get_imdsv2_token():
-    # Try with getting IMDSv2 token, fall back to IMDSv1 if can not get the token
-    token = requests.put(
-        "http://169.254.169.254/latest/api/token",
-        headers={"X-aws-ec2-metadata-token-ttl-seconds": "300"},
-        timeout=METADATA_REQUEST_TIMEOUT,
-    )
-    headers = {}
-    if token.status_code == requests.codes.ok:
-        headers["X-aws-ec2-metadata-token"] = token.content
-    return headers
 
 
 def validate_device_name(device_name):
@@ -66,47 +46,6 @@ def adapt_device_name(dev):
     return dev
 
 
-def parse_proxy_config():
-    config = configparser.RawConfigParser()
-    config.read("/etc/boto.cfg")
-    proxy_config = Config()
-    if config.has_option("Boto", "proxy") and config.has_option("Boto", "proxy_port"):
-        proxy = config.get("Boto", "proxy")
-        proxy_port = config.get("Boto", "proxy_port")
-        proxy_config = Config(proxies={"https": f"{proxy}:{proxy_port}"})
-    return proxy_config
-
-
-def get_device_volume_id(ec2, dev, instance_id):
-    # Poll for blockdevicemapping
-    devices = ec2.describe_instance_attribute(InstanceId=instance_id, Attribute="blockDeviceMapping").get(
-        "BlockDeviceMappings"
-    )
-    dev_map = dict((d.get("DeviceName"), d) for d in devices)
-    loop_count = 0
-    while dev not in dev_map:
-        if loop_count == 36:
-            syslog.syslog(f"Dev {dev} did not appears in 180 seconds.")
-            sys.exit(1)
-        syslog.syslog(f"Looking for dev {dev} in dev_map {dev_map}")
-        time.sleep(5)
-        devices = ec2.describe_instance_attribute(InstanceId=instance_id, Attribute="blockDeviceMapping").get(
-            "BlockDeviceMappings"
-        )
-        dev_map = dict((d.get("DeviceName"), d) for d in devices)
-        loop_count += 1
-
-    return dev_map.get(dev).get("Ebs").get("VolumeId")
-
-
-def get_metadata_value(token, metadata_path):
-    return requests.get(
-        metadata_path,
-        headers=token,
-        timeout=METADATA_REQUEST_TIMEOUT,
-    ).text
-
-
 def main():
     syslog.syslog("Starting ec2_dev_2_volid.py script")
     try:
@@ -115,29 +54,14 @@ def main():
         syslog.syslog(f"Input block device is {dev}")
     except IndexError:
         syslog.syslog(syslog.LOG_ERR, "Provide block device i.e. xvdf")
-
     dev = adapt_device_name(dev)
-
-    token = get_imdsv2_token()
-
-    instance_id = get_metadata_value(token, "http://169.254.169.254/latest/meta-data/instance-id")
-
-    region = get_metadata_value(token, "http://169.254.169.254/latest/meta-data/placement/availability-zone")
-    region = region[:-1]
-
-    proxy_config = parse_proxy_config()
-
-    # Configure the AWS CA bundle.
-    # In US isolated regions the dedicated CA bundle will be used.
-    # In any other region, the default bundle will be used (None stands for the default settings).
-    # Note: We want to apply a more general solution that applies to every region,
-    # but for the time being this is enough to support US isolated regions without
-    # impacting the other ones.
-    ca_bundle = f"/etc/pki/{region}/certs/ca-bundle.pem" if region.startswith("us-iso") else None
-
-    ec2 = boto3.client("ec2", region_name=region, config=proxy_config, verify=ca_bundle)
-
-    volume_id = get_device_volume_id(ec2, dev, instance_id)
+    mapping_file_path = "/dev/disk/by-ebs-volumeid/parallelcluster_dev_id_mapping"
+    if os.path.isfile(mapping_file_path):
+        with open(mapping_file_path, "r", encoding="utf-8") as mapping_file:
+            mapping = json.load(mapping_file)
+    else:
+        mapping = {}
+    volume_id = mapping.get(dev)
     print(volume_id)
 
 

cookbooks/aws-parallelcluster-environment/files/default/ec2_udev_rules/manageVolume.py

@@ -7,6 +7,7 @@
 
 import argparse
 import configparser
+import json
 import os
 import re
 import subprocess  # nosec B404
@@ -133,6 +134,17 @@ def attach_volume(volume_id, instance_id, ec2):
     dev = available_devices[0]
     response = ec2.attach_volume(VolumeId=volume_id, InstanceId=instance_id, Device=dev)
 
+    mapping_file_path = "/dev/disk/by-ebs-volumeid/parallelcluster_dev_id_mapping"
+    if os.path.isfile(mapping_file_path):
+        with open(mapping_file_path, "r", encoding="utf-8") as mapping_file:
+            mapping = json.load(mapping_file)
+    else:
+        mapping = {}
+    mapping[dev] = volume_id
+    os.makedirs(os.path.dirname(mapping_file_path), exist_ok=True)
+    with open(mapping_file_path, "w", encoding="utf-8") as mapping_file:
+        json.dump(mapping, mapping_file)
+
     # Poll for volume to attach
     state = response.get("State")
     delay = 5  # seconds

cookbooks/aws-parallelcluster-environment/recipes/config/imds.rb

@@ -60,8 +60,8 @@
     command "mkdir -p $(dirname #{ip6tables_rules_file}) && ip6tables-save > #{ip6tables_rules_file}"
   end
 
-  template '/etc/init.d/parallelcluster-iptables' do
-    source 'imds/parallelcluster-iptables.erb'
+  template '/usr/local/sbin/restore_tables.sh' do
+    source 'imds/restore_tables.sh.erb'
     user 'root'
     group 'root'
     mode '0744'
@@ -71,6 +71,25 @@
     )
   end
 
+  template '/usr/local/sbin/save_tables.sh' do
+    source 'imds/save_tables.sh.erb'
+    user 'root'
+    group 'root'
+    mode '0744'
+    variables(
+      iptables_rules_file: iptables_rules_file,
+      ip6tables_rules_file: ip6tables_rules_file
+    )
+  end
+
+  template '/etc/systemd/system/parallelcluster-iptables.service' do
+    source 'imds/parallelcluster-iptables.service.erb'
+    cookbook 'aws-parallelcluster-environment'
+    owner 'root'
+    group 'root'
+    mode '0644'
+  end
+
   service "parallelcluster-iptables" do
     action %i(enable start)
   end

cookbooks/aws-parallelcluster-environment/resources/cloudwatch/cloudwatch_redhat8.rb

@@ -13,7 +13,7 @@
 # See the License for the specific language governing permissions and limitations under the License.
 
 provides :cloudwatch, platform: 'redhat' do |node|
-  node['platform_version'].to_i == 8
+  node['platform_version'].to_i >= 8
 end
 
 use 'partial/_cloudwatch_common'

cookbooks/aws-parallelcluster-environment/resources/cloudwatch/cloudwatch_rocky8.rb

@@ -13,7 +13,7 @@
 # See the License for the specific language governing permissions and limitations under the License.
 
 provides :cloudwatch, platform: 'rocky' do |node|
-  node['platform_version'].to_i == 8
+  node['platform_version'].to_i >= 8
 end
 
 use 'partial/_cloudwatch_common'

cookbooks/aws-parallelcluster-environment/resources/ec2_udev_rules/ec2_udev_rules_redhat8.rb

@@ -13,7 +13,7 @@
 # This file is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, express or implied.
 # See the License for the specific language governing permissions and limitations under the License.
 provides :ec2_udev_rules, platform: 'redhat' do |node|
-  node['platform_version'].to_i == 8
+  node['platform_version'].to_i >= 8
 end
 
 unified_mode true

cookbooks/aws-parallelcluster-environment/resources/ec2_udev_rules/ec2_udev_rules_rocky8.rb

@@ -13,7 +13,7 @@
 # This file is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, express or implied.
 # See the License for the specific language governing permissions and limitations under the License.
 provides :ec2_udev_rules, platform: 'rocky' do |node|
-  node['platform_version'].to_i == 8
+  node['platform_version'].to_i >= 8
 end
 
 unified_mode true

cookbooks/aws-parallelcluster-environment/resources/ec2_udev_rules/ec2_udev_rules_ubuntu20+.rb

@@ -18,12 +18,10 @@
 
 unified_mode true
 use 'partial/_common_udev_configuration'
-use 'partial/_debian_udev_configuration'
 
 default_action :setup
 
 action :setup do
   action_create_common_udev_files
-  action_set_udev_autoreload
   action_start_ec2blk
 end

cookbooks/aws-parallelcluster-environment/resources/ec2_udev_rules/partial/_common_udev_configuration.rb

@@ -47,15 +47,6 @@
     mode '0744'
   end
 
-  cookbook_file 'ec2blkdev-init' do
-    source 'ec2_udev_rules/ec2blkdev-init'
-    cookbook 'aws-parallelcluster-environment'
-    path '/etc/init.d/ec2blkdev'
-    user 'root'
-    group 'root'
-    mode '0744'
-  end
-
   cookbook_file 'manageVolume.py' do
     source 'ec2_udev_rules/manageVolume.py'
     cookbook 'aws-parallelcluster-environment'
@@ -67,8 +58,7 @@
 end
 
 action :start_ec2blk do
-  service "ec2blkdev" do
-    supports restart: true
-    action %i(enable start)
+  execute "Refresh UdevAdmin" do
+    command "udevadm trigger --action=change --subsystem-match=block"
   end unless on_docker?
 end