Update docker support for Nemo/Nccl testing #18

amaslenn · 2024-05-17T10:30:45Z

Summary

1. Test by @amaslenn
CI
2. Test by @TaekyungHeo
2.1 Slurm command generation for a given URL

  [cmd_args.docker_image_url]
  type = "str"
  default = [URL]

$ cloudaix --mode dry-run --system_config_path ... --test_scenario_path conf/v0.6/general/test_scenario/nemo_launcher.toml

... container=**[installed_docker_image_path]**

2.1 Slurm command generation for a given path

  [cmd_args.docker_image_url]
  type = "str"
  default = [local_file_path]

$ cloudaix --mode dry-run --system_config_path ...--test_scenario_path conf/v0.6/general/test_scenario/nemo_launcher.toml

... container=[local_file_path]

—

amaslenn added 3 commits May 17, 2024 12:15

Add get_docker_image_path() for NcclTestSlurmCommandGenStrategy

48f3984

Add set_container_arg() for NeMoLauncherSlurmCommandGenStrategy

f0356b2

Extend NeMoLauncherSlurmInstallStrategy docker image support

70f2f99

amaslenn requested review from TaekyungHeo and srinivas212 May 17, 2024 10:31

artemry-nv approved these changes May 17, 2024

View reviewed changes

Merge branch 'main' into docker-path

78b0d3f

srinivas212 approved these changes May 18, 2024

View reviewed changes

srinivas212 merged commit 6dc690e into main May 18, 2024
2 checks passed

This was referenced May 18, 2024

Revert "Update docker support for Nemo/Nccl testing" #19

Closed

Fetch online container #22

Closed

srinivas212 deleted the docker-path branch May 21, 2024 06:07