Skip to content

fix(mori): unblock MORI smoke on containerized RDMA clusters - #453

Open
speriaswamy-amd wants to merge 1 commit into
mainfrom
speriaswamy/mori-docker-validation-fixes-main
Open

speriaswamy-amd wants to merge 1 commit into
mainfrom
speriaswamy/mori-docker-validation-fixes-main

Conversation

@speriaswamy-amd

Copy link
Copy Markdown
Contributor

Changes

  • docker_lib.py: optional ulimit_memlock param (MORI-only caller)
  • mori_benchmark_test.py: pass ulimit_memlock='-1' for RDMA ibv_reg_mr
  • mori_lib.py: export LD_LIBRARY_PATH from torchlib_dir (CVS-004)
  • mori_lib.py: accept test done! for concurrent_put_thread / concurrent_put_imm_thread smoke checks

Test plan

  • MORI smoke on Alola/TensorWave (concurrent_put_*, shmem)

Made with Cursor

Set memlock ulimit only for MORI container launch (ibv_reg_mr needs it;
Docker default 8MB breaks RDMA registration). Export torchlib_dir via
LD_LIBRARY_PATH in mori env script (CVS-004). Accept concurrent_put_thread/
imm_thread completion marker as smoke pass.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant