Merge pull request #358 from rhatdan/instructlab

Fixes for configuring training
containers · Apr 29, 2024 · 5c705b2 · 5c705b2
2 parents b148710 + a7cc5e8
commit 5c705b2
Show file tree

Hide file tree

Showing 5 changed files with 46 additions and 4 deletions.
diff --git a/training/amd/Containerfile b/training/amd/Containerfile
@@ -5,12 +5,24 @@ ADD rocm.repo /etc/yum.repos.d/rocm.repo
 # Enable CRB repo for CentOS. For RHEL, it should be enabled on the build host
 RUN source /etc/os-release \
     && export OS_ID="$ID" \
-    && if [ "$OS_ID" == "centos" ]; then \
-        curl -L https://dl.fedoraproject.org/pub/epel/epel-release-latest-9.noarch.rpm -o /tmp/epel-release-latest-9.noarch.rpm \
-        && rpm -ivh /tmp/epel-release-latest-9.noarch.rpm \
-        && rm -f /tmp/epel-release-latest-9.noarch.rpm \
+    && if [ "$OS_ID" == "centos" ] || [ "$OS_ID" == "rhel" ]; then \
+        rpm -ivh https://dl.fedoraproject.org/pub/epel/epel-release-latest-9.noarch.rpm \
         && dnf install -y 'dnf-command(config-manager)' \
         && crb enable ;\
     fi
 
 RUN dnf install -y rocm && dnf clean all
+
+ARG INSTRUCTLAB_IMAGE="quay.io/ai-lab/instructlab-amd:latest"
+
+# Setup /usr/lib/containers/storage as an additional store for images.
+# Remove once the base images have this set by default.
+RUN sed -i -e '/additionalimage.*/a "/usr/lib/containers/storage",' \
+        /etc/containers/storage.conf
+
+# Added for running as an OCI Container to prevent Overlay on Overlay issues.
+VOLUME /var/lib/containers
+
+# Prepull the instructlab image
+
+RUN podman pull --arch=${TARGETARCH} --root /usr/lib/containers/storage ${INSTRUCTLAB_IMAGE}
diff --git a/training/amd/Makefile b/training/amd/Makefile
@@ -11,7 +11,13 @@ CONTAINER_TOOL_EXTRA_ARGS ?=
 .PHONY: image
 image:
 	"${CONTAINER_TOOL}" build \
+	        --security-opt label=disable \
+		--cap-add SYS_ADMIN \
 		--file Containerfile \
 		--tag "${REGISTRY}/${REGISTRY_ORG}/${IMAGE_NAME}:${IMAGE_TAG}" \
 		$(FROM:%=--from=%) \
 		${CONTAINER_TOOL_EXTRA_ARGS}
+
+.PHONY: push
+push:
+	podman push "${REGISTRY}/${REGISTRY_ORG}/${IMAGE_NAME}:${IMAGE_TAG}"
diff --git a/training/intel/Makefile b/training/intel/Makefile
@@ -20,3 +20,7 @@ image:
 		$(DRIVER_VERSION:%=--build-arg DRIVER_VERSION=%) \
 		$(KERNEL_VERSION:%=--build-arg KERNEL_VERSION=%) \
 		${CONTAINER_TOOL_EXTRA_ARGS}
+
+.PHONY: push
+push:
+	podman push "${REGISTRY}/${REGISTRY_ORG}/${IMAGE_NAME}:${IMAGE_TAG}"
diff --git a/training/nvidia/Containerfile b/training/nvidia/Containerfile
@@ -122,3 +122,17 @@ RUN if [ "${TARGET_ARCH}" == "" ]; then \
     && dnf clean all \
     && ln -s /usr/lib/systemd/system/nvidia-toolkit-firstboot.service /usr/lib/systemd/system/basic.target.wants/nvidia-toolkit-firstboot.service \
     && echo "blacklist nouveau" > /etc/modprobe.d/blacklist_nouveau.conf
+
+ARG INSTRUCTLAB_IMAGE="quay.io/ai-lab/instructlab-nvidia:latest"
+
+# Setup /usr/lib/containers/storage as an additional store for images.
+# Remove once the base images have this set by default.
+RUN sed -i -e '/additionalimage.*/a "/usr/lib/containers/storage",' \
+        /etc/containers/storage.conf
+
+# Added for running as an OCI Container to prevent Overlay on Overlay issues.
+VOLUME /var/lib/containers
+
+# Prepull the instructlab image
+
+RUN podman pull --arch=${TARGETARCH} --root /usr/lib/containers/storage ${INSTRUCTLAB_IMAGE}
diff --git a/training/nvidia/Makefile b/training/nvidia/Makefile
@@ -32,6 +32,8 @@ dtk:
 		.
 image:
 	"${CONTAINER_TOOL}" build \
+	        --security-opt label=disable \
+		--cap-add SYS_ADMIN \
 		$(ARCH:%=--platform linux/%) \
 		--file Containerfile \
 		--tag "${REGISTRY}/${REGISTRY_ORG}/${IMAGE_NAME}:${IMAGE_TAG}" \
@@ -43,3 +45,7 @@ image:
 		$(CUDA_VERSION:%=--build-arg CUDA_VERSION=%) \
 		${CONTAINER_TOOL_EXTRA_ARGS} \
 		.
+
+.PHONY: push
+push:
+	podman push "${REGISTRY}/${REGISTRY_ORG}/${IMAGE_NAME}:${IMAGE_TAG}"