Skip to content

fix(mori,docker): unblock MORI smoke on containerized RDMA clusters - #451

Open
speriaswamy-amd wants to merge 1 commit into
release/cvs-0.2.0from
speriaswamy/mori-docker-validation-fixes
Open

speriaswamy-amd wants to merge 1 commit into
release/cvs-0.2.0from
speriaswamy/mori-docker-validation-fixes

Conversation

@speriaswamy-amd

Copy link
Copy Markdown
Contributor

Changes

  • docker_lib.py: add --ulimit memlock=-1 (Docker defaults to 8MB; breaks ibv_reg_mr for RDMA)
  • 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 (these binaries do not print PASSED)

Test plan

  • MORI smoke on Alola/TensorWave (concurrent_put_*, shmem)
  • Verify other Docker-launched RDMA suites still launch with unlimited memlock

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.
@speriaswamy-amd
speriaswamy-amd force-pushed the speriaswamy/mori-docker-validation-fixes branch from 108a730 to 4c858a1 Compare September 17, 2026 23:43
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.

2 participants