Optimize container images for startup (#6547)

* Optimize container images for startup

This change adjusts how to handle runner payloads to support
container builds where we keep them extracted in the filesystem.
This makes it easier to optimize the cpu/cuda vs cpu/rocm images for
size, and should result in faster startup times for container images.

* Refactor payload logic and add buildx support for faster builds

* Move payloads around

* Review comments

* Converge to buildx based helper scripts

* Use docker buildx action for release

scripts/build_darwin.sh

@@ -2,8 +2,7 @@
 
 set -e
 
-export VERSION=${VERSION:-$(git describe --tags --first-parent --abbrev=7 --long --dirty --always | sed -e "s/^v//g")}
-export GOFLAGS="'-ldflags=-w -s \"-X=github.com/ollama/ollama/version.Version=$VERSION\" \"-X=github.com/ollama/ollama/server.mode=release\"'"
+. $(dirname $0)/env.sh
 
 mkdir -p dist
 

scripts/build_docker.sh

@@ -2,76 +2,34 @@
 
 set -eu
 
-export VERSION=${VERSION:-$(git describe --tags --first-parent --abbrev=7 --long --dirty --always | sed -e "s/^v//g")}
-export GOFLAGS="'-ldflags=-w -s \"-X=github.com/ollama/ollama/version.Version=$VERSION\" \"-X=github.com/ollama/ollama/server.mode=release\"'"
-
-# We use 2 different image repositories to handle combining architecture images into multiarch manifest
-# (The ROCm image is x86 only and is not a multiarch manifest)
-# For developers, you can override the DOCKER_ORG to generate multiarch manifests
-#  DOCKER_ORG=jdoe PUSH=1 ./scripts/build_docker.sh
-DOCKER_ORG=${DOCKER_ORG:-"ollama"}
-RELEASE_IMAGE_REPO=${RELEASE_IMAGE_REPO:-"${DOCKER_ORG}/release"}
-FINAL_IMAGE_REPO=${FINAL_IMAGE_REPO:-"${DOCKER_ORG}/ollama"}
-
-BUILD_ARCH=${BUILD_ARCH:-"amd64 arm64"}
+. $(dirname $0)/env.sh
 
 # Set PUSH to a non-empty string to trigger push instead of load
 PUSH=${PUSH:-""}
 
-# In CI mode, we break things down
-OLLAMA_SKIP_MANIFEST_CREATE=${OLLAMA_SKIP_MANIFEST_CREATE:-""}
-OLLAMA_SKIP_IMAGE_BUILD=${OLLAMA_SKIP_IMAGE_BUILD:-""}
-
 if [ -z "${PUSH}" ] ; then
+    echo "Building ${FINAL_IMAGE_REPO}:$VERSION locally.  set PUSH=1 to push"
     LOAD_OR_PUSH="--load"
 else
-    echo "Will be pushing ${RELEASE_IMAGE_REPO}:$VERSION for ${BUILD_ARCH}"
+    echo "Will be pushing ${FINAL_IMAGE_REPO}:$VERSION"
     LOAD_OR_PUSH="--push"
 fi
 
-if [ -z "${OLLAMA_SKIP_IMAGE_BUILD}" ]; then
-    for TARGETARCH in ${BUILD_ARCH}; do
-        docker build \
-            ${LOAD_OR_PUSH} \
-            --platform=linux/${TARGETARCH} \
-            --build-arg=VERSION \
-            --build-arg=GOFLAGS \
-            -f Dockerfile \
-            -t ${RELEASE_IMAGE_REPO}:$VERSION-${TARGETARCH} \
-            .
-    done
+docker buildx build \
+    ${LOAD_OR_PUSH} \
+    --platform=${PLATFORM} \
+    ${OLLAMA_COMMON_BUILD_ARGS} \
+    -f Dockerfile \
+    -t ${FINAL_IMAGE_REPO}:$VERSION \
+    .
 
-    if echo ${BUILD_ARCH} | grep "amd64" > /dev/null; then
-        docker build \
-            ${LOAD_OR_PUSH} \
-            --platform=linux/amd64 \
-            --build-arg=VERSION \
-            --build-arg=GOFLAGS \
-            --target runtime-rocm \
-            -f Dockerfile \
-            -t ${RELEASE_IMAGE_REPO}:$VERSION-rocm \
-            .
-    fi
-fi
-
-if [ -z "${OLLAMA_SKIP_MANIFEST_CREATE}" ]; then
-    if [ -n "${PUSH}" ]; then
-        docker manifest create ${FINAL_IMAGE_REPO}:$VERSION \
-            ${RELEASE_IMAGE_REPO}:$VERSION-amd64 \
-            ${RELEASE_IMAGE_REPO}:$VERSION-arm64
-        docker manifest push ${FINAL_IMAGE_REPO}:$VERSION
-
-        # For symmetry, tag/push the rocm image
-        if [ "${RELEASE_IMAGE_REPO}" != "${FINAL_IMAGE_REPO}" ]; then
-            echo "Tagging and pushing rocm image"
-            docker pull ${RELEASE_IMAGE_REPO}:$VERSION-rocm
-            docker tag ${RELEASE_IMAGE_REPO}:$VERSION-rocm ${FINAL_IMAGE_REPO}:$VERSION-rocm
-            docker push ${FINAL_IMAGE_REPO}:$VERSION-rocm
-        fi
-    else
-        echo "Skipping manifest generation when not pushing images are available locally as "
-        echo "  ${RELEASE_IMAGE_REPO}:$VERSION-amd64"
-        echo "  ${RELEASE_IMAGE_REPO}:$VERSION-arm64"
-        echo "  ${RELEASE_IMAGE_REPO}:$VERSION-rocm"
-    fi
-fi
+if echo $PLATFORM | grep "amd64" > /dev/null; then
+    docker buildx build \
+        ${LOAD_OR_PUSH} \
+        --platform=linux/amd64 \
+        ${OLLAMA_COMMON_BUILD_ARGS} \
+        --target runtime-rocm \
+        -f Dockerfile \
+        -t ${FINAL_IMAGE_REPO}:$VERSION-rocm \
+        .
+fi

scripts/build_linux.sh

@@ -1,37 +1,29 @@
 #!/bin/sh
+#
+# Mac ARM users, rosetta can be flaky, so to use a remote x86 builder
+#
+# docker context create amd64 --docker host=ssh://mybuildhost
+# docker buildx create --name mybuilder amd64 --platform linux/amd64
+# docker buildx create --name mybuilder --append desktop-linux --platform linux/arm64
+# docker buildx use mybuilder
+
 
 set -eu
 
-export VERSION=${VERSION:-$(git describe --tags --first-parent --abbrev=7 --long --dirty --always | sed -e "s/^v//g")}
-export GOFLAGS="'-ldflags=-w -s \"-X=github.com/ollama/ollama/version.Version=$VERSION\" \"-X=github.com/ollama/ollama/server.mode=release\"'"
-GZIP=$(which pigz 2>/dev/null || echo "gzip")
+. $(dirname $0)/env.sh
 
-BUILD_ARCH=${BUILD_ARCH:-"amd64 arm64"}
-export AMDGPU_TARGETS=${AMDGPU_TARGETS:=""}
 mkdir -p dist
 
-for TARGETARCH in ${BUILD_ARCH}; do
-    docker build \
-        --platform=linux/$TARGETARCH \
-        --build-arg=GOFLAGS \
-        --build-arg=CGO_CFLAGS \
-        --build-arg=OLLAMA_CUSTOM_CPU_DEFS \
-        --build-arg=AMDGPU_TARGETS \
-        --target build-$TARGETARCH \
+docker buildx build \
+        --output type=local,dest=./dist/ \
+        --platform=${PLATFORM} \
+        ${OLLAMA_COMMON_BUILD_ARGS} \
+        --target dist \
         -f Dockerfile \
-        -t builder:$TARGETARCH \
         .
-    docker create --platform linux/$TARGETARCH --name builder-$TARGETARCH builder:$TARGETARCH
-    rm -rf ./dist/linux-$TARGETARCH
-    docker cp builder-$TARGETARCH:/go/src/github.com/ollama/ollama/dist/linux-$TARGETARCH ./dist
-    if echo ${TARGETARCH} | grep "amd64" > /dev/null; then
-        docker cp builder-$TARGETARCH:/go/src/github.com/ollama/ollama/dist/linux-$TARGETARCH-rocm ./dist
-    fi
-    docker rm builder-$TARGETARCH
-    echo "Compressing final linux bundle..."
-    rm -f ./dist/ollama-linux-$TARGETARCH.tgz
-    (cd dist/linux-$TARGETARCH && tar cf - . | ${GZIP} --best > ../ollama-linux-$TARGETARCH.tgz )
-    if [ -d dist/linux-$TARGETARCH-rocm ]; then
-        (cd dist/linux-$TARGETARCH-rocm && tar cf - . | ${GZIP} --best > ../ollama-linux-$TARGETARCH-rocm.tgz )
-    fi
-done
+
+# buildx behavior changes for single vs. multiplatform
+if echo $PLATFORM | grep "," > /dev/null ; then 
+        mv -f ./dist/linux_*64/ollama* ./dist/
+        rmdir ./dist/linux_*64
+fi

scripts/env.sh

@@ -0,0 +1,14 @@
+# Common environment setup across build*.sh scripts
+
+export VERSION=${VERSION:-$(git describe --tags --first-parent --abbrev=7 --long --dirty --always | sed -e "s/^v//g")}
+export GOFLAGS="'-ldflags=-w -s \"-X=github.com/ollama/ollama/version.Version=$VERSION\" \"-X=github.com/ollama/ollama/server.mode=release\"'"
+# TODO - consider `docker buildx ls --format=json` to autodiscover platform capability
+PLATFORM=${PLATFORM:-"linux/arm64,linux/amd64"}
+DOCKER_ORG=${DOCKER_ORG:-"ollama"}
+RELEASE_IMAGE_REPO=${RELEASE_IMAGE_REPO:-"${DOCKER_ORG}/release"}
+FINAL_IMAGE_REPO=${FINAL_IMAGE_REPO:-"${DOCKER_ORG}/ollama"}
+OLLAMA_COMMON_BUILD_ARGS="--build-arg=VERSION --build-arg=GOFLAGS --build-arg=OLLAMA_CUSTOM_CPU_DEFS --build-arg=AMDGPU_TARGETS"
+
+echo "Building Ollama"
+echo "VERSION=$VERSION"
+echo "PLATFORM=$PLATFORM"