[ARVADOS] updated: 2.1.0-1744-gec5a52d35
Git user
git at public.arvados.org
Fri Dec 17 14:50:57 UTC 2021
Summary of changes:
tools/compute-images/scripts/base.sh | 47 ++++++++++++++++++++++++++----------
1 file changed, 34 insertions(+), 13 deletions(-)
via ec5a52d3551e558e6df50c50e94118d84b0cde08 (commit)
from fcbfddb10723cb876a1c83e883ce3bfb4f6a2565 (commit)
Those revisions listed above that are new to this repository have
not appeared on any other notification email; so we list those
revisions in full, below.
commit ec5a52d3551e558e6df50c50e94118d84b0cde08
Author: Ward Vandewege <ward at curii.com>
Date: Fri Dec 17 09:49:33 2021 -0500
18325: when using Nvidia GPUs on Debian 10, we need a newer Docker. Also
fix use of the $SUDO variable in the base packer script and add a
final call to `apt-get clean` to it.
Arvados-DCO-1.1-Signed-off-by: Ward Vandewege <ward at curii.com>
diff --git a/tools/compute-images/scripts/base.sh b/tools/compute-images/scripts/base.sh
index 973a49310..e70eccd0c 100644
--- a/tools/compute-images/scripts/base.sh
+++ b/tools/compute-images/scripts/base.sh
@@ -153,31 +153,52 @@ $SUDO chown root:root /etc/cloud/cloud.cfg.d/07_compute_arvados_dispatch_cloud.c
if [ "$NVIDIA_GPU_SUPPORT" == "1" ]; then
DIST=$(. /etc/os-release; echo $ID$VERSION_ID)
# We need a kernel and matching headers
- $sudo apt-get -y install linux-image-cloud-amd64 linux-headers-cloud-amd64
+ $SUDO apt-get -y install linux-image-cloud-amd64 linux-headers-cloud-amd64
# Install CUDA
- $sudo apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/$DIST/x86_64/7fa2af80.pub
- $sudo apt-get -y install software-properties-common
- $sudo add-apt-repository "deb https://developer.download.nvidia.com/compute/cuda/repos/$DIST/x86_64/ /"
- $sudo add-apt-repository contrib
- $sudo apt-get update
- $sudo apt-get -y install cuda
+ $SUDO apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/$DIST/x86_64/7fa2af80.pub
+ $SUDO apt-get -y install software-properties-common
+ $SUDO add-apt-repository "deb https://developer.download.nvidia.com/compute/cuda/repos/$DIST/x86_64/ /"
+ $SUDO add-apt-repository contrib
+ $SUDO apt-get update
+ $SUDO apt-get -y install cuda
# Install libnvidia-container, the tooling for Docker/Singularity
curl -s -L https://nvidia.github.io/libnvidia-container/gpgkey | \
- $sudo apt-key add -
+ $SUDO apt-key add -
if [ "$DIST" == "debian11" ]; then
# As of 2021-12-16 libnvidia-container and friends are only available for
# Debian 10, not yet Debian 11. Install experimental rc1 package as per this
# workaround:
# https://github.com/NVIDIA/nvidia-docker/issues/1549#issuecomment-989670662
curl -s -L https://nvidia.github.io/libnvidia-container/debian10/libnvidia-container.list | \
- $sudo tee /etc/apt/sources.list.d/libnvidia-container.list
- $sudo sed -i -e '/experimental/ s/^#//g' /etc/apt/sources.list.d/libnvidia-container.list
+ $SUDO tee /etc/apt/sources.list.d/libnvidia-container.list
+ $SUDO sed -i -e '/experimental/ s/^#//g' /etc/apt/sources.list.d/libnvidia-container.list
else
curl -s -L https://nvidia.github.io/libnvidia-container/$DIST/libnvidia-container.list | \
- $sudo tee /etc/apt/sources.list.d/libnvidia-container.list
+ $SUDO tee /etc/apt/sources.list.d/libnvidia-container.list
fi
- $sudo apt-get update
- $sudo apt-get -y install libnvidia-container1 libnvidia-container-tools nvidia-container-toolkit
+
+ if [ "$DIST" == "debian10" ]; then
+ # Debian 10 comes with Docker 18.xx, we need 19.03 or later
+ curl -fsSL https://download.docker.com/linux/debian/gpg | $SUDO gpg --dearmor -o /usr/share/keyrings/docker-archive-keyring.gpg
+ echo deb [arch=amd64 signed-by=/usr/share/keyrings/docker-archive-keyring.gpg] https://download.docker.com/linux/debian/ buster stable | \
+ $SUDO tee /etc/apt/sources.list.d/docker.list
+ $SUDO apt-get update
+ $SUDO apt-get -yq --no-install-recommends install docker-ce=5:19.03.15~3-0~debian-buster
+
+ $SUDO sed "s/ExecStart=\(.*\)/ExecStart=\1 --default-ulimit nofile=10000:10000 ${SET_RESOLVER}/g" \
+ /lib/systemd/system/docker.service \
+ > /etc/systemd/system/docker.service
+
+ $SUDO systemctl daemon-reload
+
+ # docker should not start on boot: we restart it inside /usr/local/bin/ensure-encrypted-partitions.sh,
+ # and the BootProbeCommand might be "docker ps -q"
+ $SUDO systemctl disable docker
+ fi
+ $SUDO apt-get update
+ $SUDO apt-get -y install libnvidia-container1 libnvidia-container-tools nvidia-container-toolkit
fi
+
+$SUDO apt-get clean
-----------------------------------------------------------------------
hooks/post-receive
--
More information about the arvados-commits
mailing list