WeichenXu123 commented on code in PR #39267:
URL: https://github.com/apache/spark/pull/39267#discussion_r1072300627
##########
python/pyspark/ml/torch/distributor.py:
##########
@@ -325,8 +329,15 @@ def _create_torchrun_command(
torchrun_args = ["--standalone", "--nnodes=1"]
processes_per_node = num_processes
else:
- pass
- # TODO(SPARK-41592): Handle distributed training
+ master_addr, master_port = os.environ["MASTER_ADDR"],
os.environ["MASTER_PORT"]
+ node_rank = os.environ["RANK"]
+ torchrun_args = [
+ f"--nnodes={num_processes}",
+ f"--node_rank={node_rank}",
+ f"--rdzv_endpoint={master_addr}:{master_port}",
+ "--rdzv_id=0",
+ ] # TODO: setup random ID that is gleaned from env variables
+ processes_per_node = 1
Review Comment:
We don't need setting `preexec_fn=sigterm_on_parent_death` when executing
`torch_run_process_wrapper` subprocess
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
