aws-parallelcluster 3.16.0__tar.gz → 3.16.2__tar.gz

{aws_parallelcluster-3.16.0 → aws_parallelcluster-3.16.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: aws-parallelcluster
-Version: 3.16.0
+Version: 3.16.2
 Summary: AWS ParallelCluster is an AWS supported Open Source cluster management tool to deploy and manage HPC clusters in the AWS cloud.
 Home-page: https://github.com/aws/aws-parallelcluster
 Author: Amazon Web Services
@@ -17,6 +17,7 @@ Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: 3.13
+Classifier: Programming Language :: Python :: 3.14
 Classifier: Topic :: Scientific/Engineering
 Classifier: License :: OSI Approved :: Apache Software License
 Requires-Python: >=3.9

{aws_parallelcluster-3.16.0 → aws_parallelcluster-3.16.2}/setup.py

@@ -20,7 +20,7 @@ def readme():
         return f.read()
 
 
-VERSION = "3.16.0"
+VERSION = "3.16.2"
 CDK_VERSION = "1.164"
 REQUIRES = [
     "setuptools",
@@ -93,6 +93,7 @@ setup(
         "Programming Language :: Python :: 3.11",
         "Programming Language :: Python :: 3.12",
         "Programming Language :: Python :: 3.13",
+        "Programming Language :: Python :: 3.14",
         "Topic :: Scientific/Engineering",
         "License :: OSI Approved :: Apache Software License",
     ],

{aws_parallelcluster-3.16.0 → aws_parallelcluster-3.16.2}/src/aws_parallelcluster.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: aws-parallelcluster
-Version: 3.16.0
+Version: 3.16.2
 Summary: AWS ParallelCluster is an AWS supported Open Source cluster management tool to deploy and manage HPC clusters in the AWS cloud.
 Home-page: https://github.com/aws/aws-parallelcluster
 Author: Amazon Web Services
@@ -17,6 +17,7 @@ Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: 3.13
+Classifier: Programming Language :: Python :: 3.14
 Classifier: Topic :: Scientific/Engineering
 Classifier: License :: OSI Approved :: Apache Software License
 Requires-Python: >=3.9

{aws_parallelcluster-3.16.0 → aws_parallelcluster-3.16.2}/src/pcluster/api/openapi/openapi.yaml

@@ -2,7 +2,7 @@ openapi: 3.0.2
 info:
   description: ParallelCluster API
   title: ParallelCluster
-  version: 3.16.0
+  version: 3.16.2
 servers:
 - url: /
 # override: auth is defined the the API GW level

{aws_parallelcluster-3.16.0 → aws_parallelcluster-3.16.2}/src/pcluster/aws/aws_resources.py

@@ -391,7 +391,7 @@ class FsxStorageInfo:
     @property
     def network_interface_ids(self):
         """Return network interface ids of the file system."""
-        return self.file_storage_info.get("NetworkInterfaceIds")
+        return self.file_storage_info.get("NetworkInterfaceIds", [])
 
     @property
     def subnet_ids(self):

{aws_parallelcluster-3.16.0 → aws_parallelcluster-3.16.2}/src/pcluster/aws/logs.py

@@ -49,7 +49,7 @@ class LogsClient(Boto3Client):
 
     @AWSExceptionHandler.handle_client_exception
     def filter_log_events(self, log_group_name, start_time=None, end_time=None, log_stream_name_prefix=None):
-        """Return the list of events included in a specific time window for a given group name."""
+        """Return the FilterLogEvents response for a specific time window for a given group name."""
         kwargs = {"logGroupName": log_group_name, "limit": 1}
         if start_time:
             kwargs["startTime"] = start_time
@@ -57,7 +57,7 @@ class LogsClient(Boto3Client):
             kwargs["endTime"] = end_time
         if log_stream_name_prefix:
             kwargs["logStreamNamePrefix"] = log_stream_name_prefix
-        return self._client.filter_log_events(**kwargs).get("events")
+        return self._client.filter_log_events(**kwargs)
 
     @AWSExceptionHandler.handle_client_exception
     def get_log_events(
@@ -121,6 +121,28 @@ class LogsClient(Boto3Client):
             )
         return tasks[0].get("status").get("code")
 
+    @AWSExceptionHandler.handle_client_exception
+    def get_active_export_tasks(self):
+        """Return the CloudWatch Logs export tasks that are currently active (PENDING or RUNNING).
+
+        CloudWatch Logs allows only one active export task per account at a time, so this is used to give
+        a clear "an export is already running, please wait" message instead of a raw LimitExceededException
+        when a second export is attempted (e.g. after the user interrupts and reruns the command).
+        """
+        active_tasks = []
+        for status_code in ("PENDING", "PENDING_CANCEL", "RUNNING"):
+            next_token = None
+            while True:
+                kwargs = {"statusCode": status_code}
+                if next_token:
+                    kwargs["nextToken"] = next_token
+                response = self._client.describe_export_tasks(**kwargs)
+                active_tasks.extend(response.get("exportTasks", []))
+                next_token = response.get("nextToken")
+                if not next_token:
+                    break
+        return active_tasks
+
     @AWSExceptionHandler.handle_client_exception
     def describe_log_streams(self, log_group_name, log_stream_name_prefix=None, next_token=None):
         """Return a list of log streams in the given log group, filtered by the given prefix."""

{aws_parallelcluster-3.16.0 → aws_parallelcluster-3.16.2}/src/pcluster/config/cluster_config.py

@@ -30,10 +30,7 @@ from pcluster.config.common import (
     DefaultUserHomeType,
 )
 from pcluster.config.common import Imds as TopLevelImds
-from pcluster.config.common import (
-    Resource,
-    SharedStorageType,
-)
+from pcluster.config.common import Resource, SharedStorageType
 from pcluster.constants import (
     CIDR_ALL_IPS,
     COMPUTE_INSTANCE_INFO_TIMEOUT,
@@ -67,6 +64,7 @@ from pcluster.constants import (
     Feature,
 )
 from pcluster.utils import (
+    create_hash_suffix,
     get_attr,
     get_partition,
     get_resource_name_from_resource_arn,
@@ -96,6 +94,7 @@ from pcluster.validators.cluster_validators import (
     InstanceArchitectureCompatibilityValidator,
     IntelHpcArchitectureValidator,
     IntelHpcOsValidator,
+    ManagedFsxCrossAzValidator,
     ManagedFsxMultiAzValidator,
     MaxCountValidator,
     MixedSecurityGroupOverwriteValidator,
@@ -1622,6 +1621,16 @@ class BaseClusterConfig(Resource):
         self.deployment_settings = deployment_settings
         self.managed_head_node_security_group = None
         self.managed_compute_security_group = None
+        # Dictionary {mount target logical id: subnet id} with an entry for each EFS mount target of the deployed stack,
+        # e.g. {"EFS2395b171bc347024MTuswest2a": "subnet-12345678"}.
+        # It is set during update to keep each mount target in its current subnet: EFS allows a single mount target
+        # per AZ, so moving a mount target to another subnet of the same AZ would make the stack update fail.
+        self.existing_efs_mount_target_subnets = {}
+        # Dictionary {file system logical id: subnet ids} with an entry for each FSx file system of the deployed stack,
+        # e.g. {"FSX3c9d7e6a8c2b1f05": ["subnet-12345678"]}.
+        # It is set during update to keep each managed FSx in its current subnet: changing the subnet of an FSx
+        # replaces the file system, deleting the existing one and its data.
+        self.existing_fsx_file_system_subnets = {}
         self.instance_types_data_version = ""
 
     def _register_validators(self, context: ValidatorContext = None):  # noqa: D102 #pylint: disable=unused-argument
@@ -1800,6 +1809,18 @@ class BaseClusterConfig(Resource):
                     queues=queues,
                     fsx_az_list=storage.file_system_availability_zones,
                 )
+            if isinstance(storage, SharedFsxLustre) and not storage.is_unmanaged:
+                # During update, the existing file system is kept in its subnets, see existing_fsx_file_system_subnets
+                existing_subnet_ids = self.existing_fsx_file_system_subnets.get(
+                    f"{SharedStorageType.FSX.name}{create_hash_suffix(storage.name)}"
+                )
+                if existing_subnet_ids:
+                    self._register_validator(
+                        ManagedFsxCrossAzValidator,
+                        fsx_name=storage.name,
+                        fsx_subnet_ids=existing_subnet_ids,
+                        compute_subnet_ids=compute_subnet_ids,
+                    )
             if isinstance(storage, SharedEbs):
                 ebs_volumes.append(storage)
 

{aws_parallelcluster-3.16.0 → aws_parallelcluster-3.16.2}/src/pcluster/config/common.py

@@ -20,6 +20,7 @@ from abc import ABC, abstractmethod
 from enum import Enum
 from typing import List, Set
 
+from pcluster.utils import event_loop
 from pcluster.validators.common import AsyncValidator, FailureLevel, ValidationResult, Validator, ValidatorContext
 from pcluster.validators.dev_settings_validators import ExtraChefAttributesValidator
 from pcluster.validators.iam_validators import AdditionalIamPolicyValidator
@@ -208,11 +209,9 @@ class Resource:
         # if they are taking too long to execute for a resource and its children since the use of
         # get_async_timed_validator_type_for allows to decorate only on single validator at a time and
         # does not cascade to child resources
-        return list(
-            itertools.chain.from_iterable(
-                asyncio.get_event_loop().run_until_complete(asyncio.gather(*self._validation_futures))
-            )
-        )
+        with event_loop() as loop:
+            validation_results = loop.run_until_complete(asyncio.gather(*self._validation_futures))
+        return list(itertools.chain.from_iterable(validation_results))
 
     def _nested_resources(self):
         nested_resources = []
@@ -312,6 +311,20 @@ class BaseTag(Resource):
         self.value = Resource.init_param(value)
 
 
+def merge_tags(*tag_lists):
+    """
+    Merge lists of Tag objects, deduplicating by key.
+
+    Later lists take precedence over earlier ones on duplicate keys, mirroring the tag precedence
+    applied at launch time (cluster < queue < compute resource). ``None`` lists are skipped.
+    """
+    merged = {}
+    for tags in tag_lists:
+        for tag in tags or []:
+            merged[tag.key] = tag
+    return list(merged.values())
+
+
 class AdditionalIamPolicy(Resource):
     """Represent the Additional IAM Policy configuration."""
 

{aws_parallelcluster-3.16.0 → aws_parallelcluster-3.16.2}/src/pcluster/constants.py

@@ -39,7 +39,6 @@ SUPPORTED_OSES_FOR_P6E_GB200 = list(set(SUPPORTED_OSES) - set(UNSUPPORTED_OSES_F
 UNSUPPORTED_OSES_FOR_P6_B300 = []
 SUPPORTED_OSES_FOR_P6_B300 = list(set(SUPPORTED_OSES) - set(UNSUPPORTED_OSES_FOR_P6_B300))
 UNSUPPORTED_OSES_FOR_DCV = []
-UNSUPPORTED_OSES_FOR_NON_GPU_DCV = ["rocky9", "rhel9"]
 UNSUPPORTED_ARM_OSES_FOR_DCV = []
 UNSUPPORTED_OSES_FOR_LUSTRE = []
 DELETE_POLICY = "Delete"
@@ -135,7 +134,7 @@ EBS_VOLUME_TYPE_IOPS_DEFAULT = {
 }
 EBS_VOLUME_SIZE_DEFAULT = 40
 # This value must be greater than or equal to the AMI volume size
-EBS_ROOT_VOLUME_SIZE_DEFAULT = 45
+EBS_ROOT_VOLUME_SIZE_DEFAULT = 50
 EBS_VOLUME_TYPE_DEFAULT = "gp3"
 
 DEFAULT_MAX_COUNT = 10
@@ -156,8 +155,8 @@ MAX_NEW_STORAGE_COUNT = {"efs": 1, "fsx": 1, "raid": 1}
 MAX_EXISTING_STORAGE_COUNT = {"efs": 20, "fsx": 20, "raid": 0}
 
 COOKBOOK_PACKAGES_VERSIONS = {
-    "parallelcluster": "3.16.0",
-    "cookbook": "aws-parallelcluster-cookbook-3.16.0",
+    "parallelcluster": "3.16.2",
+    "cookbook": "aws-parallelcluster-cookbook-3.16.2",
     "chef": "19.3.14",
     "ami": "dev",
 }
@@ -362,28 +361,3 @@ ULTRASERVER_CAPACITY_BLOCK_ALLOWED_SIZE_DICT = {
 }
 # Capacity Block states that are considered inactive (cannot check health status)
 CAPACITY_BLOCK_INACTIVE_STATES = ["scheduled", "payment-pending", "assessing", "delayed"]
-
-# Older generation instance types
-EXCLUDED_INSTANCE_TYPE_PREFIXES = (
-    "m1",
-    "m2",
-    "m3",
-    "m4",
-    "t1",
-    "t2",
-    "c1",
-    "c3",
-    "c4",
-    "r3",
-    "r4",
-    "x1",
-    "x1e",
-    "d2",
-    "h1",
-    "i2",
-    "i3",
-    "f1",
-    "g3",
-    "p2",
-    "p3",
-)

{aws_parallelcluster-3.16.0 → aws_parallelcluster-3.16.2}/src/pcluster/models/cluster.py

@@ -55,6 +55,7 @@ from pcluster.models.common import (
     create_logs_archive,
     export_stack_events,
     parse_config,
+    sanitize_path_component,
     upload_archive,
 )
 from pcluster.models.compute_fleet_status_manager import ComputeFleetStatus, ComputeFleetStatusManager
@@ -487,6 +488,8 @@ class Cluster:
             if context.during_update:
                 config.managed_head_node_security_group = self.stack.get_resource_physical_id("HeadNodeSecurityGroup")
                 config.managed_compute_security_group = self.stack.get_resource_physical_id("ComputeSecurityGroup")
+                config.existing_efs_mount_target_subnets = self.stack.efs_mount_target_subnets
+                config.existing_fsx_file_system_subnets = self.stack.fsx_file_system_subnets
 
             validation_failures = config.validate(validator_suppressors, context)
             if any(f.level.value >= FailureLevel(validation_failure_level).value for f in validation_failures):
@@ -970,8 +973,9 @@ class Cluster:
 
         try:
             with tempfile.TemporaryDirectory() as output_tempdir:
-                # Create root folder for the archive
-                archive_name = f"{self.name}-logs-{datetime.now().strftime('%Y%m%d%H%M')}"
+                # Create root folder for the archive. Sanitize the identifier: it may be an ARN (deleted
+                # clusters can only be referenced by ARN), whose '/' and ':' would break local paths.
+                archive_name = f"{sanitize_path_component(self.name)}-logs-{datetime.now().strftime('%Y%m%d%H%M')}"
                 root_archive_dir = os.path.join(output_tempdir, archive_name)
                 os.makedirs(root_archive_dir, exist_ok=True)
 
@@ -997,8 +1001,11 @@ class Cluster:
                         {self.name},
                     )
 
-                # Get stack events and write them into a file
-                stack_events_file = os.path.join(root_archive_dir, self._stack_events_stream_name)
+                # Get stack events and write them into a file (sanitize: the stream name embeds the
+                # stack identifier, which may be an ARN containing path separators).
+                stack_events_file = os.path.join(
+                    root_archive_dir, sanitize_path_component(self._stack_events_stream_name)
+                )
                 export_stack_events(self.stack_name, stack_events_file)
 
                 archive_path = create_logs_archive(root_archive_dir, output_file)

{aws_parallelcluster-3.16.0 → aws_parallelcluster-3.16.2}/src/pcluster/models/cluster_resources.py

@@ -18,6 +18,7 @@ from pcluster.aws.aws_api import AWSApi
 from pcluster.aws.aws_resources import InstanceInfo, StackInfo
 from pcluster.constants import CW_LOGS_CFN_PARAM_NAME, OS_MAPPING, PCLUSTER_NODE_TYPE_TAG, PCLUSTER_VERSION_TAG
 from pcluster.models.common import FiltersParserError, LogGroupTimeFiltersParser, get_all_stack_events
+from pcluster.utils import yaml_load
 
 
 class ClusterStack(StackInfo):
@@ -72,6 +73,26 @@ class ClusterStack(StackInfo):
         """Return the original official AMI."""
         return self._get_param("OfficialAmi")
 
+    @property
+    def efs_mount_target_subnets(self):
+        """Return the subnet of each EFS mount target in the stack template, keyed by the mount target logical id."""
+        resources = yaml_load(AWSApi.instance().cfn.get_stack_template(self.name)).get("Resources", {})
+        return {
+            logical_id: resource["Properties"]["SubnetId"]
+            for logical_id, resource in resources.items()
+            if resource.get("Type") == "AWS::EFS::MountTarget"
+        }
+
+    @property
+    def fsx_file_system_subnets(self):
+        """Return the subnets of each FSx file system in the stack template, keyed by the file system logical id."""
+        resources = yaml_load(AWSApi.instance().cfn.get_stack_template(self.name)).get("Resources", {})
+        return {
+            logical_id: resource["Properties"]["SubnetIds"]
+            for logical_id, resource in resources.items()
+            if resource.get("Type") == "AWS::FSx::FileSystem"
+        }
+
     def delete(self):
         """Delete stack."""
         AWSApi.instance().cfn.delete_stack(self.name)