Skip to content

Releases: aws/aws-parallelcluster-node

AWS ParallelCluster v3.0.1

27 Oct 14:25
Compare
Choose a tag to compare

We're excited to announce the release of AWS ParallelCluster Node 3.0.1

This is associated with AWS ParallelCluster v3.0.1

3.0.1

CHANGES

  • There were no changes for this version.

AWS ParallelCluster v3.0.0

10 Sep 15:53
Compare
Choose a tag to compare

We're excited to announce the release of AWS ParallelCluster Node 3.0.0

This is associated with AWS ParallelCluster v3.0.0

3.0.0

ENHANCEMENTS

  • Implement scaling protection mechanism with Slurm scheduler: compute fleet is automatically set to 'PROTECTED' state
    in case recurrent failures are encountered when provisioning nodes.
  • Implement computemgtd self-termination via shutdown command instead of calling TerminateInstances.

CHANGES

  • Drop support for SGE and Torque schedulers.
  • Use tags prefix parallelcluster: when describing EC2 instances.
  • Run Slurm command scontrol with sudo because clustermgtd is executed as cluster admin user (not root).

AWS ParallelCluster v2.11.2

26 Aug 17:03
Compare
Choose a tag to compare

We're excited to announce the release of AWS ParallelCluster Node 2.11.2

This is associated with AWS ParallelCluster v2.11.2

2.11.2

BUG FIXES

  • Slurm: fix issue that prevented powering-up nodes to be correctly reset after a stop and start of the cluster.

AWS ParallelCluster v2.11.1

23 Jul 23:52
Compare
Choose a tag to compare

We're excited to announce the release of AWS ParallelCluster Node 2.11.1

This is associated with AWS ParallelCluster v2.11.1

CHANGES

  • There were no notable changes for this version.

AWS ParallelCluster v2.11.0

01 Jul 04:02
Compare
Choose a tag to compare

We're excited to announce the release of AWS ParallelCluster Node 2.11.0

This is associated with AWS ParallelCluster v2.11.0

ENHANCEMENTS

  • SGE: always use shortname as hostname filter with qstat. This will make nodewatcher more robust when using custom DHCP option, where the full hostname seen by SGE might differ from the hostname returned from EC2 metadata(local-hostname).
  • Transition from IMDSv1 to IMDSv2.
  • Have computemgtd reuse last available daemon configuration when the new one cannot be loaded.
  • Use methods with timeouts to read NFS shared files, which will prevent computemgtd from hanging when NFS filesystems are not available.

BUG FIXES

  • Fix a bug that caused clustermgtd to not immediately replace instances with failed status check that are in replacement process.

AWS ParallelCluster v2.10.4

15 May 17:06
Compare
Choose a tag to compare

We're excited to announce the release of AWS ParallelCluster Node 2.10.4

This is associated with AWS ParallelCluster v2.10.4

CHANGES

  • There were no notable changes for this version.

AWS ParallelCluster v2.10.3

18 Mar 22:06
Compare
Choose a tag to compare

We're excited to announce the release of AWS ParallelCluster Node 2.10.3

This is associated with AWS ParallelCluster v2.10.3

CHANGES

  • There were no notable changes for this version.

AWS ParallelCluster v2.10.2

02 Mar 16:33
Compare
Choose a tag to compare

We're excited to announce the release of AWS ParallelCluster Node 2.10.2

This is associated with AWS ParallelCluster v2.10.2

CHANGES

  • There were no notable changes for this version.

AWS ParallelCluster v2.10.1

22 Dec 23:16
Compare
Choose a tag to compare

We're excited to announce the release of AWS ParallelCluster Node 2.10.1

This is associated with AWS ParallelCluster v2.10.1

ENHANCEMENTS

  • Improve error handling in slurm plugin processes when clustermgtd is down.

CHANGES

  • Increase max attempts when retrying on Route53 API call failures.

AWS ParallelCluster v2.10.0

18 Nov 16:21
Compare
Choose a tag to compare

We're excited to announce the release of AWS ParallelCluster Node 2.10.0

This is associated with AWS ParallelCluster v2.10.0

ENHANCEMENTS

  • Add new all_or_nothing_batch configuration parameter for slurm_resume script. When True, slurm_resume will
    succeed only if all the instances required by all the pending jobs in Slurm will be available.

CHANGES

  • CentOS 6 is no longer supported.
  • Optimize retrieval of nodes info from Slurm scheduler.
  • Improve retrieval of instance type info by using DescribeInstanceType API.
  • Increase timeout from 10 to 30 seconds when clustermgtd and computemgtd daemons invoke Slurm commands.

BUG FIXES

  • Retrieve the right number of compute instance slots when instance type is updated.
  • Fix a bug that was causing clustermgtd and computemgtd sleep interval to be incorrectly computed when
    system timezone is not set to UTC.