Skip to content

DAOSGCP-176 Merge develop to main#84

Merged
mark-olson merged 11 commits intomainfrom
develop
May 18, 2023
Merged

DAOSGCP-176 Merge develop to main#84
mark-olson merged 11 commits intomainfrom
develop

Conversation

@mark-olson
Copy link
Copy Markdown
Contributor

@mark-olson mark-olson commented May 18, 2023

  • Changes necessary to support the HPC Rocky Linux 8 image
  • Changes necessary to support DAOS > v2.2.0
  • Removal of MIGs in DAOS Terraform modules
  • Disable hyperthreading. Configure 1 thread per vCPU.
  • Refactor IO500 terraform example
  • Fix broken link in Pre-Deployment guide

Signed-off-by: Mark Olson 115657904+mark-olson@users.noreply.github.com

mark-olson and others added 5 commits December 12, 2022 10:38
Merge from main to develop after v0.3.0
Signed-off-by: Artem Gayardo <artemgm@google.com>
DAOSGCP-161 Fix setting the --region parameter
DAOS will now be deployed with images based on Rocky Linux 8

Changes:

Reduced total number of files in images/ directory
Replaced shell scripts with Ansible playbooks
Renamed build_images.sh to build.sh
Removed hard-coded strings in Packer template
Simplified Google Cloud Build file (packer_cloudbuild.yml)
Added support for running Packer locally instead of running in Cloud Build

Signed-off-by: Mark Olson <115657904+mark-olson@users.noreply.github.com>
Remove Managed Instance Group (MIG) provisioning from Terrform modules.

For large deployments provisioning instances with google_compute_instance vs
google_compute_instance_group_manager reduces provisioning time.

This change was provided by Margaret Lawson.

Signed-off-by: Margaret Lawson <mlawsonca@google.com>
Signed-off-by: Mark Olson <115657904+mark-olson@users.noreply.github.com>
@mark-olson mark-olson requested a review from ravalsam May 18, 2023 15:51
Copy link
Copy Markdown

@ravalsam ravalsam left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks ok as it's merge from dev to main branch,

@mark-olson mark-olson force-pushed the develop branch 2 times, most recently from fc99585 to 6c6a7d5 Compare May 18, 2023 16:17
Refactored terraform/examples/io500

Changes starting with '[ML]' were either contributed directly or suggested by Margaret Lawson.

- Wrapper scripts moved to bin/ directory
- Client scripts moved to client_scripts/ directory
- Scripts in images/ directory updated to wrap new images/build.sh script
- Do not use terraform/examples/daos_cluster for io500 example. terraform/examples/io500 has its own terraform configuration now.
- Removed the use of TF_VARS_* environment variables. Now uses a terraform.tfvars file that is generated based on the config/*.sh file that is used by the start.sh script.
- Set default log level to INFO
- Removed references to MIGs
- [ML] Restart clients after clean
- [ML] SERVER_OS_FAMILY and CLIENT_OS_FAMILY adjusted to include the word "DAOS"
- [ML] Do not use preemptible instances for IO500

Signed-off-by: Mark Olson <115657904+mark-olson@users.noreply.github.com>
Signed-off-by: Mark Olson <115657904+mark-olson@users.noreply.github.com>
* DAOSGCP-172 Change base image and tuned profile

Changed base image to 'hpc-rocky-linux-8'
Changed tuned profile to 'network-throughput'
Fixed issue with --force option in io500 example
Fixed issue with LOG_LEVEL in io500 example
Remove unnecessary boot parameters in tune.yml ansible file

Signed-off-by: Mark Olson <115657904+mark-olson@users.noreply.github.com>
Changes tested with DAOS v2.2.0 and v2.3.107

- dmg command line changes for >v2.2.0
- daos command line changes for >v2.2.0
- removed environment file from build_io500_images.sh so that vars pass through from config/*
- added ability to set chunk size in io500 configurations
- fix cert permissions to work with v2.2.0 and v2.3.*
- added log.debug.show_vars function for debugging

Signed-off-by: Mark Olson <115657904+mark-olson@users.noreply.github.com>
Reducing the number of threads that run on each physical CPU core can help
improve the performance of workloads that are highly-parallel.

Signed-off-by: Mark Olson <115657904+mark-olson@users.noreply.github.com>
Fixed broken link in predeployment guide.

Signed-off-by: Mark Olson <115657904+mark-olson@users.noreply.github.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Development

Successfully merging this pull request may close these issues.

3 participants