aws/aws-parallelcluster v3.15.0

AWS ParallelCluster v3.15.0

on GitHub

We're excited to announce the release of AWS ParallelCluster 3.15.0

Upgrade

How to upgrade?

sudo pip install --upgrade aws-parallelcluster

3.15.0

ENHANCEMENTS

• Add support for p6-b300 instances for all OSs except AL2.
• Replace cfn-hup in compute nodes with systemd timer to support in place updates in order to improve performance for tightly coupled worloads at scale.
  This new mechanism relies on shared storage to sync updates between the head node and compute nodes.
• Disable dnf-makecache.timer to improve performance for tightly coupled worloads on RHEL/Rocky at scale.
• Support updates of Tags during cluster-updates.
• Add LaunchTemplateOverrides to cluster config to allow network interfaces to be customized by overriding the launch template of a compute resource.
  • This overrides the parallelcluster default using a shallow merge.
• Add alarm on missing clustermgtd heartbeat.

CHANGES

• When EFA is enabled, ParallelCluster now configures network interfaces as interface and efa-only instead of the combined efa type. NCI-0 is configured with an interface ENI for IP connectivity plus an efa-only ENI for EFA fabric. Secondary cards are configured with an efa-only only ENI. This reduces IP address consumption from one per network card to one per instance. EFA performance is unchanged. Customers who need the legacy efa behavior can set DevSettings: EfaInterfaceType: efa.
• Reduce transient build-image failures in RHEL and Rocky caused by out-of-sync repo mirrors by resetting metadata upon retry.
• Always start clustermgtd on cluster update and compute fleet status update failure, regardless the failure condition.
• Improve resiliency of the cluster update rollback workflow.
• Upgrade Slurm to version 25.11.4 (from 24.11.7).
• Upgrade Pmix to 5.0.10 (from 5.0.6).
• Upgrade EFA installer to 1.47.0 (from 1.44.0).
  • Efa-driver: efa-3.0.0
  • Efa-config: efa-config-1.18-1
  • Efa-profile: efa-profile-1.7-1
  • Libfabric-aws: libfabric-aws-2.4.0-1
  • Rdma-core: rdma-core-61.0-1
  • Open MPI: openmpi40-aws-4.1.7-2 and openmpi50-aws-5.0.9
• Upgrade NVIDIA driver to version 580.105.08 (from 570.172.08) for all OSs except Amazon Linux 2.
• Upgrade GDRCopy to version 2.5.2 (from 2.4.4).
• Upgrade DCV to version 2025.0-20103 (from 2024.0-19030).
• Upgrade CUDA Toolkit to version 13.0.2 (from 12.8.1) for all OSs except Amazon Linux 2.
• Upgrade NVIDIA Fabric manager to 580.105.08 for all OSs except Amazon Linux 2.
• Upgrade Python to 3.14.2 (from 3.12.11) for all OSs except Amazon Linux 2.
• Upgrade aws-cfn-bootstrap to version 2.0-38 (from 2.0-33).
• Upgrade DCGM to version 4.5.1 (from 4.4.1) for all OSs except Amazon Linux 2.
• Upgrade Munge to version 0.5.17 (from 0.5.16) for all OSs except Amazon Linux 2.
• Upgrade mysql-community-client to version 8.4.8 (from 8.0.39) for all OSs except Amazon Linux 2.
• Upgrade Intel MPI Library to 2021.17.2 (from 2021.16.0).
• Upgrade Cinc Client to version 18.8.54 (from 18.7.10).
• Upgrade amazon-efs-utils to version 2.4.0 (from v2.1.0) for Amazon Linux AMI's.
• Upgrade jmespath to ~=1.0 (from ~=0.10).
• Upgrade tabulate to <=0.9.0 (from <=0.8.10).
• Add a validator to warn in case in-place updates have been disabled (via DevSettings) on compute and login nodes.

BUG FIXES

• Fix LoginNodes NLB not being publicly accessible when using public subnet with implicit main route table association.
  See #7173
• Fix a failure when creating a cluster with GPU instances and with DCV enabled but without internet access.
• Fix an issue where cluster creation would intermittently fail due to eventual consistency when the head/compute/login nodes share the same security group.
• Fix build-image failure during ubuntu-desktop installation on a Ubuntu parent image with outdated OS packages.
• Fix validation of HeadNode/LocalStorage. This configuration parameter does not support updates.
• Fix validator PlacementGroupCapacityReservationValidator to accept capacity reservations with cross-account placement group.
• Fix the CloudWatch agent configuration to ensure proper parsing of timestamps across all log files.
• Fix logging configuration to capture all Slurm health check events (updating log level from WARNING to INFO to prevent missing log entries).
• Improve cluster update resiliency by ensuring the update does not fail on nodes completing the bootstrap during the update.
• Prevent cluster update failure recovery process from running on AWS Batch clusters. This recovery mechanism should only execute on Slurm clusters.

DEPRECATIONS

• The LoginNodes/Pools/Ssh/KeyName configuration parameter, deprecated since 3.14.0, is no longer supported.
• This is the last ParallelCluster release supporting Amazon Linux 2, as Amazon Linux 2 will reach end of support on June 30, 2026.
• This is the last ParallelCluster release supporting AWS Batch CLI. Starting with v3.16.0, ParallelCluster will no longer support AWS Batch as a scheduler.