aws-eda-slurm-cluster v2.3.3

What's Changed

• Update config files and fix errors found in testing new configs.
• Clean up ansible-lint errors and warnings.
• Paginate describe_instances when creating head node a record.
• Add default MungeKeySecret.
• Increase timeout for ssm command that configures submitters so slurm has time to compile.
• Force slurm to be rebuilt for submitters of all os distributions even if they match the os of the cluster.
• Paginate describe_instances in UpdateHeadNode lambda
• Add check for min memory of 4 GB for slurm controller
• Update documentation

New Features

• Feature #207: Add --RESEnvironmentName to the installer to ease integration with Research and Engineering Studio (RES).

Bug Fixes

• Bug #203: slurm_zfs.yml doesn't work
  slurm zfsyml doesnt work by @cartalla in #214
• Bug #206: Default head node instance type for arm cluster is incorrect
  Set default head not instance type based on cluster architecture.

Full Changelog: v2.3.2...v2.3.3

aws-eda-slurm-cluster v2.3.2

What's Changed

Bug Fixes

• Bug #200: Getting EC2 instance info fails
  Ignore pricing lists for capacity blocks by @cartalla in #201
• Bug #202: Changing controller instance type doesn't cause cluster to be updated
  Update cluster when config file changes by @cartalla in #205

Full Changelog: v2.3.1...v2.3.2

aws-eda-slurm-cluster v2.3.1

New Features

No new features in this release,

Bug Fixes

• Bug #189 - [BUG] Deployment fails if RESEnvironmentName not configured
• Bug #190 - [BUG] Handle case where ParallelCluster database stack isn't ready yet
• Bug #191 - [BUG] Deployment fails if submitter security groups not configured
• #193 - [DOCS] Baseline setup instructions from a clean AMI
• Bug #196 - [BUG] install.sh fails if CDK bootstrap stack doesn't exist
• Bug #197 - [BUG] Allow creation of stack in default VPC with only public subnets

aws-eda-slurm-cluster v2.3.0

New Features

No new features in this release, but the way that the config file is passed the Lambda that creates the cluster is significant enough
to justify a new minor version.

Bug Fixes

• Bug #179 - Stack update fails because json payload too large
• Bug #184 - Build files contain CFN tokens until stack is updated
• Bug #184 - Rocky 8 AMI builds fail
• Bug #186 - ParallelCluster deployment fails if AMI doesn't support instance type

aws-eda-slurm-cluster v2.2.0

New Features

• Feature #183 - Add support for ParallelCluster 3.8.0
  Adds support for custom munge key
  Adds support for Rocky Linux 8

aws-eda-slurm-cluster v2.1.0

New Features

• Feature #177 - Build slurm for distribution of the submitter
• Feature #181 - Automate configuration of RES cluster manager and submitters

Some pretty significant updates to automate RES configuration and deconfiguration when the cluster is deleted.

Bug Fixes

• Bug #160 - Reduce IAM permissions for CreateParallelCluster lambda
• Bug #178 - Playbook updates don't get updated correctly
• Bug #180 - Cluster not updated when scripts saved in s3 are updated

aws-eda-slurm-cluster v2.0.3

New Features

• Feature #170 - Integrate with RES

Bug Fixes

• Bug #167 - Modulefile not created
• Bug #172 - When deconfiguring submitter, fstab entry isn't removed

aws-eda-slurm-cluster v2.0.2

New Features

• Feature #164 - Make ParallelCluster 3.7.1 the default version

Bug Fixes

Bug #165 - Slurm not mounting after desktop reboot

aws-eda-slurm-cluster v2.0.1

Bug Fixes

• #161 - Fix create_user_groups.py when /usr/sbin not in path

aws-eda-slurm-cluster v2.0.0

Remove support for the legacy cluster and only support ParallelCluster based Slurm clusters.

The legacy cluster is still available for reference of the v1 branch, but will no longer be developed or supported.