The GRPO training process hangs for multi-node training. #3934
Comments
|
hangs for me too @hjh0119 can u help |
|
It might be due to OOM. BTW, the external vLLM server is now supported, which means you can use TP in async mode. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
I launched GRPO training script on a multi-node cluster (2 nodes with 8xH100 connected with infiniband), but the
training process always hangs
on 5th step without any warning or error message. I was using "Colocate mode" of GRPO since Async mode doesn't
support tensor or pipeline parallelism techniques for inferencing,
which is crucial for big models like
meta-llama/Llama-3.3-70B-Instruct.nvidia-smishows that all GPUs are utilizingmemory but the actual compute utilization was 0%. Could it be that infiniband is not configured properly?
What I've tried
datasets and enabled dataset shuffling, but it still was hanging on 5th step
--sleep-levelargumentsping private_ipwas working between nodesHow to reproduce?
Launch command 1:
Launch command 2:
Debugging info
pip freeze
printenv
SHELL=/bin/bash PWD=/home/ubuntu/ms-swift LOGNAME=ubuntu XDG_SESSION_TYPE=tty MOTD_SHOWN=pam HOME=/home/ubuntu LANG=C.UTF-8 LS_COLORS=rs=0:di=01;34:ln=01;36:mh=00:pi=40;33:so=01;35:do=01;35:bd=40;33;01:cd=40;33;01:or=40;31;01:mi=00:su=37;41:sg=30;43:ca=30;41:tw=30;42:ow=34;42:st=37;44:ex=01;32:*.tar=01;31:*.tgz=01;31:*.arc=01;31:*.arj=01;31:*.taz=01;31:*.lha=01;31:*.lz4=01;31:*.lzh=01;31:*.lzma=01;31:*.tlz=01;31:*.txz=01;31:*.tzo=01;31:*.t7z=01;31:*.zip=01;31:*.z=01;31:*.dz=01;31:*.gz=01;31:*.lrz=01;31:*.lz=01;31:*.lzo=01;31:*.xz=01;31:*.zst=01;31:*.tzst=01;31:*.bz2=01;31:*.bz=01;31:*.tbz=01;31:*.tbz2=01;31:*.tz=01;31:*.deb=01;31:*.rpm=01;31:*.jar=01;31:*.war=01;31:*.ear=01;31:*.sar=01;31:*.rar=01;31:*.alz=01;31:*.ace=01;31:*.zoo=01;31:*.cpio=01;31:*.7z=01;31:*.rz=01;31:*.cab=01;31:*.wim=01;31:*.swm=01;31:*.dwm=01;31:*.esd=01;31:*.jpg=01;35:*.jpeg=01;35:*.mjpg=01;35:*.mjpeg=01;35:*.gif=01;35:*.bmp=01;35:*.pbm=01;35:*.pgm=01;35:*.ppm=01;35:*.tga=01;35:*.xbm=01;35:*.xpm=01;35:*.tif=01;35:*.tiff=01;35:*.png=01;35:*.svg=01;35:*.svgz=01;35:*.mng=01;35:*.pcx=01;35:*.mov=01;35:*.mpg=01;35:*.mpeg=01;35:*.m2v=01;35:*.mkv=01;35:*.webm=01;35:*.webp=01;35:*.ogm=01;35:*.mp4=01;35:*.m4v=01;35:*.mp4v=01;35:*.vob=01;35:*.qt=01;35:*.nuv=01;35:*.wmv=01;35:*.asf=01;35:*.rm=01;35:*.rmvb=01;35:*.flc=01;35:*.avi=01;35:*.fli=01;35:*.flv=01;35:*.gl=01;35:*.dl=01;35:*.xcf=01;35:*.xwd=01;35:*.yuv=01;35:*.cgm=01;35:*.emf=01;35:*.ogv=01;35:*.ogx=01;35:*.aac=00;36:*.au=00;36:*.flac=00;36:*.m4a=00;36:*.mid=00;36:*.midi=00;36:*.mka=00;36:*.mp3=00;36:*.mpc=00;36:*.ogg=00;36:*.ra=00;36:*.wav=00;36:*.oga=00;36:*.opus=00;36:*.spx=00;36:*.xspf=00;36: VIRTUAL_ENV=/home/ubuntu/envs/grpo SSH_CONNECTION=78.109.65.154 43175 10.15.24.25 22 LESSCLOSE=/usr/bin/lesspipe %s %s XDG_SESSION_CLASS=user TERM=xterm-256color LESSOPEN=| /usr/bin/lesspipe %s USER=ubuntu SHLVL=1 XDG_SESSION_ID=80 VIRTUAL_ENV_PROMPT=(grpo) XDG_RUNTIME_DIR=/run/user/1000 PS1=(grpo) \[\e]0;\u@\h: \w\a\]${debian_chroot:+($debian_chroot)}\[\033[01;32m\]\u@\h\[\033[00m\]:\[\033[01;34m\]\w\[\033[00m\]\$ SSH_CLIENT=78.109.65.154 43175 22 XDG_DATA_DIRS=/usr/local/share:/usr/share:/var/lib/snapd/desktop PATH=/home/ubuntu/envs/grpo/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin DBUS_SESSION_BUS_ADDRESS=unix:path=/run/user/1000/bus SSH_TTY=/dev/pts/0 OLDPWD=/home/ubuntu _=/usr/bin/printenvifconfig on node 1
ifconfig on node 2
NCCL logs
The text was updated successfully, but these errors were encountered: