update

spawnerdockerfile/Dockerfile.ngc

@@ -37,9 +37,9 @@ apt-get install -y flex tk ethtool libpci3 libltdl-dev bison lsof tcl libelf1 pc
 mkdir -p ${STAGE_DIR}
 wget -q -O - http://www.mellanox.com/downloads/ofed/MLNX_OFED-${MLNX_OFED_VERSION}/MLNX_OFED_LINUX-${MLNX_OFED_VERSION}-ubuntu22.04-x86_64.tgz | tar xzf -
 cd ${STAGE_DIR}/MLNX_OFED_LINUX-${MLNX_OFED_VERSION}-ubuntu22.04-x86_64
-./mlnxofedinstall --user-space-only --without-fw-update --all -q 
+./mlnxofedinstall --user-space-only --without-fw-update --all -q > /tmp/mlnxofedinstall.log 2>&1
 cd ${STAGE_DIR} 
-rm -rf ${STAGE_DIR}/MLNX_OFED_LINUX-${MLNX_OFED_VERSION}-ubuntu22.04-x86_64*
+# rm -rf ${STAGE_DIR}/MLNX_OFED_LINUX-${MLNX_OFED_VERSION}-ubuntu22.04-x86_64*
 EOT
 
 ARG NV_PEER_MEM_VERSION="1.2"
@@ -107,6 +107,7 @@ EOT
 
 RUN <<EOT
 #!/bin/bash
+apt install -y cmake gcc g++ libssl-dev
 git clone https://github.com/microsoft/DeepSpeed-Kernels.git ${STAGE_DIR}/DeepSpeed-Kernels
 cd ${STAGE_DIR}/DeepSpeed-Kernels
 python -m pip install -v .
@@ -160,6 +161,9 @@ python -m pip install black~=23.1 ruff==0.1.5 diffusers>=0.17.0
 python -m pip install --no-deps git+https://github.com/huggingface/optimum.git#egg=optimum[diffusers,quality]
 EOT
 
+
+# SSH daemon port inside container cannot conflict with host OS port
+ENV SSH_PORT=2222
 RUN <<EOT
 #!/bin/bash
 # 项目目录中的定义通常会覆盖用户家目录中的定义
@@ -174,6 +178,8 @@ CUDA_HOME=${CUDA_HOME}
 LD_LIBRARY_PATH=${LD_LIBRARY_PATH}
 EOF
 unset https_proxy http_proxy
+cat /etc/ssh/sshd_config > ${STAGE_DIR}/sshd_config && \
+sed "0,/^Port 22/s//Port ${SSH_PORT}/" ${STAGE_DIR}/sshd_config > /etc/ssh/sshd_config
 EOT
 
 CMD ["/usr/sbin/sshd", "-D"]

spawnerdockerfile/docker-compose_ngc.yml

@@ -45,7 +45,7 @@ services:
     # tty: true
     # stdin_open: true
     restart: unless-stopped
-    image: hotwa/notebook:ngc
+    image: quay.io/hotwa/ngc:latest
     privileged: true
     ipc: host
     network_mode: host