This file is used to list changes made in each version of the AWS ParallelCluster cookbook.
ENHANCEMENTS
- FSx Lustre - add support in Amazon Linux
CHANGES
- Slurm - upgrade to version 18.08.6.2
- Slurm - declare nodes in separate config file and use FUTURE for dummy nodes
- Slurm - set
ReturnToService=1
in scheduler config in order to recover instances that were initially marked as down due to a transient issue. - NVIDIA - update drivers to version 418.56
- CUDA - update toolkit to version 10.0
- Increase default EBS volume size from 15GB to 17GB
- Add
LocalHostname
toCOMPUTE_READY
events - Pin
future
,retrying
andsix
packages in Ubuntu 14.04 - Add
stackname
andmax_queue_size
to sqswatcher configuration
ENHANCEMENTS
FSx Lustre
: add support for FSx Lustre in Centos7. In case of custom AMI, FSx Lustre is only supported with Centos 7.5 and Centos 7.6.
CHANGES
SGE
: allow users to force job deletionCentos7
: use official AMI as the base image when building ParallelCluster AMI
BUG FIXES
Torque
: wait for scheduler initialization before completing compute node setupEBS
: fix block device conversion to correctly attach ebs nvme volumesPacker
: retrieveaws-cfn-bootstrap-latest
package fromcn-north-1
in all China regions.RAID
: automatically prepend/
toraid_shared_dir
when not specified in cluster configEBS
: remove extra leading/
when mounting EBS volumes
TOOLING
- Add a script to simplify cookbook package upload when using
custom_chef_cookbook
option
- China Regions, cn-north-1 and cn-northwest-1 support
- EFS support
- RAID support
- Fix issue with jq on ubuntu1404 and centos6. Now using version 1.4
- Fix dependency issue with AWS CLI package on ubuntu1404
- Rename CfnCluster to AWS ParallelCluster
- Support multiple EBS Volumes
- Add AWS Batch as a supported scheduler
- Support Custom AMI's
- Add
scaledown_idletime
to nodewatcher config - Add cookbook recipes for jobwatcher
- Remove publish_pending scripts
- Set SGE Accounting summary to be true, this reports a single accounting record for a mpi job
- Add option to disable ganglia
extra_json = { "cfncluster" : { "ganglia_enabled" : "no" } }
- Fix bug that prevented c5d/m5d instances from working
- Set CPU as a consumable resource in slurm config
Major new features/updates:
- Added parameter to specify custom cfncluster-node package
Bug fixes/minor improvements:
- Fixed poise-python dependecy issue
- Poll on EBS Volume attachment status
- Added more info on failure of pre and post install files
- Fixed SLURM cron job to publish pending metric
Major new features/updates:
- Updated to latest cfncluster-node 1.4.3
Major new features/updates:
- Updated to Amazon Linux 2017.09.1
- Applied patches to Ubuntu 16.04
- Applied patches to Ubuntu 14.04
- Updated to Centos 7.4
- Upgraded Centos 6 AMI
- Updated to Nvidia driver 384
- Updated to CUDA 9
- Updated to latest cfncluster-node 1.4.2
Bug fixes/minor improvements:
- Added support for NVMe-based instance store
- Fixed ganglia plotting issue on ubuntu
- Fixed slow SLURM scaling times on systemd platforms.
- Relicensed to Apache License 2.0
- Updated to Amazon Linux 2017.03
- Pulled in latest cookbook dependencies
- Removed Openlava support
- Dougal Ballantyne
- Updated to Chef 12.8.1
- Updated Openlava to 3.1.3
- Updated SGE to 8.1.9
- Updated cfncluster-node to 1.1.0
- Added slots to compute-ready script
- Updated cookbook dependencies
- Dougal Ballantyne - Updated to Amazon Linux 2015.09.2 for base AMI
- Dougal Ballantyne
- Fix Ganglia rebuild on 2nd run
- Update to cfncluster-node==1.0.1
- Dougal Ballantyne - 1.0.0 release of cookbook matching 1.0.0 release of cfncluster.
- Dougal Ballantyne - Initial release of cfncluster-cookbooks