Skip to content

AWS ParallelCluster v3.16.0

Latest
Latest

Choose a tag to compare

View all tags
@gmarciani gmarciani released this 20 Aug 23:48
v3.16.0
d60f292
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

We're excited to announce the release of AWS ParallelCluster 3.16.0

Upgrade

How to upgrade?

sudo pip install --upgrade aws-parallelcluster

ENHANCEMENTS

  • Add the pcluster-diag diagnostics tool to ParallelCluster AMIs for running on-demand diagnostic checks on cluster nodes.
  • Improve resilience of cluster creation by retrying EBS volume attachment on transient IMDS connectivity failures.
  • Further reduce transient build-image failures on RHEL and Rocky caused by out-of-sync repo mirrors by resetting metadata upon retry.
  • Improve resilience of cluster updates on login nodes by reusing the head-node-driven orchestration already in place on
    compute nodes, removing the dependency on cfn-hup and cfn-init.
  • Move all ParallelCluster-managed bootstrap files off /tmp into a dedicated /opt/parallelcluster/tmp
    directory, so that image builds, cluster creations, and updates work on custom AMIs that mount /tmp with noexec.
    Image builds work on the custom AMIs only if GDRcopy installation is skipped.
  • Disable fwupd-refresh.timer to improve performance for tightly coupled workloads at scale.

CHANGES

  • The validator ClusterNameValidator now limits cluster names to 40 characters when using ExternalSlurmdbd,
    consistent with the existing limit for Database. This prevents runtime failures caused by MySQL's table name length limit.
  • The validator MultiNetworkInterfacesInstancesValidator now also covers single-network-card instances with EFA enabled, which are launched with multiple network interfaces and therefore cannot be auto-assigned a public IP.
  • The CLI now requires the additional permission tag:GetResources, which it uses to resolve login nodes load balancer ARNs by tags (see the related bugfix).
  • Add support for Python 3.13 in the pcluster CLI.
  • Enforce NFSv4-only on the ParallelCluster-managed NFS server (head node). The NFSv3 client stack (rpcbind, rpc-statd, lockd)
    is unchanged, so cluster nodes can still mount external NFSv3 servers.
  • Change the default NFS lock manager (lockd) port from 32768 to 4045, to avoid collisions with the Linux ephemeral port range (32768–60999), which can cause sporadic mount failures.
    This only affects nodes that mount an external NFSv3 server; all ParallelCluster-managed storage is mounted over NFSv4 and is unaffected.
    If you restrict NFS ports in a firewall, you must open TCP/UDP 4045 or NFSv3 mounts will fail.
  • Install the NVIDIA driver, CUDA toolkit, Fabric Manager, NVLSM, and IMEX from the distribution package manager using NVIDIA local repo packages instead of the run file installers.
  • On RHEL-family OSes, install the Xorg driver for DCV GPU acceleration and disable Wayland so that GDM always starts Xorg.
    This is now required after switching NVIDIA driver installation to local repo packages; previously it was needed only on Ubuntu.
  • Upgrade NVIDIA driver, Fabric Manager, and IMEX to version 595.71.05 (from 580.126.20).
  • Upgrade CUDA Toolkit to version 13.2.2 (from 13.0.2).
  • Upgrade DCGM to version 4.6.0 (from 4.5.1).
  • In GPU Health Check, skip DCGM diagnostics when NVIDIA MIG is enabled because dcgmi diag does not support MIG.
  • Upgrade Slurm to version 25.11.6 (from 25.11.4).
  • Upgrade EFA installer to 1.49.0 (from 1.47.0).
    • Efa-driver: efa-3.1.0
    • 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-63.0-1
    • Open MPI: openmpi40-aws-4.1.7-3 and openmpi50-aws-5.0.9-11
  • Upgrade mysql-community-client to version 8.4.10 (from 8.4.8).
  • Install the aws-parallelcluster-node package from S3 in all regions instead of PyPI, to support air-gapped and proxied environments.
  • Install amazon-efs-utils from the official EFS endpoint instead of building from source.
    Users must now allowlist the CloudFront domain amazon-efs-utils.aws.com in their proxy/egress configuration
    when using build-image with isolated subnets.
  • Upgrade Cinc Client to version 19.3.14 (from 18.8.54).
  • Upgrade GDRCopy to version 2.6 (from 2.5.2).
  • Upgrade PMIx to version 5.0.11 (from 5.0.10).
  • Upgrade Enroot to version 4.2.1 (from 3.4.1).
  • Upgrade Pyxis to version 0.24.0 (from 0.20.0).
  • Upgrade stunnel to version 5.78 (from 5.67).
  • Upgrade Python to version 3.14.6 (from 3.14.2).
  • Upgrade Intel MPI to version 2021.18.0.749 (from 2021.17.2.94).
  • Upgrade Arm Performance Libraries (ArmPL) to version 26.01.1 (from 24.10).
  • Upgrade aws-cfn-bootstrap to version 2.0-40 (from 2.0-38).
  • Upgrade third-party cookbook dependencies:
    • line-5.0.0 (from line-4.5.21)
    • nfs-5.1.6 (from nfs-5.1.5)
    • openssh-2.11.17 (from openssh-2.11.14)
    • yum-8.0.0 (from yum-7.4.20)
    • yum-epel-5.0.9 (from yum-epel-5.0.8)
  • Raise the maximum allowed value of the FSx for Lustre AutomaticBackupRetentionDays parameter from 35 to 90, matching the
    Amazon FSx service limit.

BUG FIXES

  • Fix cluster creation failure when login nodes are used and CloudWatch logging is disabled.
  • Fix sporadic S3 bucket creation failure (for buckets named parallelcluster-*-v1-do-not-delete) when multiple
    create-cluster commands run simultaneously in the same region.
  • Fix cluster creation failure caused by Slurm accounting bootstrap failing when ClusterName is overridden
    via custom Slurm settings or when the cluster name contains uppercase letters.
  • Remove the deprecated AccountingStorageUser parameter from the Slurm configuration, which was causing harmless error messages.
  • Fix DCV configuration by letting the DCV server decide the display encoders for the instance type.
  • Fix DCV prerequisite installation potentially blocking on interactive prompts by exporting DEBIAN_FRONTEND=noninteractive to child processes.
  • Fix Xdcv segfault caused by DCV attempting GL initialization when GPU acceleration is not supported.
  • Fix slurmrestd failing to start on AL2023 because the http-parser library was not discoverable by the dynamic linker.
  • Fix compute node bootstrap hanging without a clear error when the compute subnet cannot reach DynamoDB.
  • Fix clustermgtd failing to detect compute node bootstrap timeouts, which prevented the cluster from entering protected mode.
  • Fix race condition in load balancer lookup by using tag:GetResources to resolve load balancer ARNs by tags,
    which previously caused cluster update failures on clusters with login nodes.
  • Fail pcluster build-image early when the downloaded cookbook version does not match the ParallelCluster CLI version.
  • Fix login nodes not mounting /opt/parallelcluster/shared when EFS is used as the internal shared storage type.
  • Fix SELinux not being disabled on RHEL-family OSes (kernels >= 6.4) due to a deprecated mechanism being silently
    ignored by newer kernels.
  • Fix build-image failure by skipping installation of fabric-manager, gdrcopy, and dcgm if the parent image already has a version installed.
  • Fix an issue where compute nodes are replaced when launching a large number of nodes due to eventual consistency.
  • Fix a race condition where a fleet start was not enough to exit protected mode.
    This race condition is triggered when a fleet stop is issued on a fleet transitioning into protected mode while some of its nodes are powering up.
  • Fix an issue where static nodes in a maintenance reservation enter a terminate/relaunch loop when being replaced.
  • Fix cluster creation and update failures caused by IAM policy tag conditions that required the parallelcluster:cluster-name
    tag to be the only tag on a request rather than merely present. This affected clusters with login nodes as well as stack
    creation and updates when AWS or CloudFormation attached additional tags to the request.
  • Prevent a path traversal when downloading exported cluster and image logs, so a CloudWatch log stream name containing ../
    segments can no longer cause files to be written outside the export directory.
  • Return a clear error instead of an unexpected fatal exception when the configured AWS profile does not exist.
  • Cache getPolicy API calls to reduce validation time of clusters that have additional IAM policies.

DEPRECATIONS

  • Amazon Linux 2 is no longer supported.
  • AWS Batch as a scheduler is no longer supported.
Assets 2
Loading