fix rdzv_id

rdzv_id may be not equal in multi node setup using RANDOM
pytorch · May 6, 2024 · d23693a · d23693a
1 parent e48aaeb
commit d23693a
Showing 1 changed file with 3 additions and 2 deletions.
diff --git a/distributed/ddp-tutorial-series/slurm/sbatch_run.sh b/distributed/ddp-tutorial-series/slurm/sbatch_run.sh
@@ -14,10 +14,11 @@ head_node_ip=$(srun --nodes=1 --ntasks=1 -w "$head_node" hostname --ip-address)
 echo Node IP: $head_node_ip
 export LOGLEVEL=INFO
 
+job_id=2024
 srun torchrun \
 --nnodes 4 \
 --nproc_per_node 1 \
---rdzv_id $RANDOM \
+--rdzv_id ${jobid} \
 --rdzv_backend c10d \
 --rdzv_endpoint $head_node_ip:29500 \
-/shared/examples/multinode_torchrun.py 50 10
+/shared/examples/multinode_torchrun.py 50 10