Merge branch 'main' into topic/06-05-feat_add_agent_count_in_scaling_…

…group_resolver

.editorconfig

@@ -16,6 +16,7 @@ indent_style = space
 indent_size = 4
 
 [*.md]
+max_line_length = 0
 trim_trailing_whitespace = false
 
 [*.rst]

.github/workflows/alembic-head-check.yml

@@ -7,7 +7,7 @@ on:
 
 jobs:
   check-multiple-heads:
-    if: ${{ contains(github.event.pull_request.labels.*.name, 'require:db-migration') && github.event.pull_request.merged == false }}
+    if: ${{ contains(github.event.pull_request.labels.*.name, 'require:db-migration') && !contains(fromJSON('["flow:merge-queue", "flow:hotfix"]'), github.event.label.name) && github.event.pull_request.merged == false }}
     runs-on: ubuntu-latest
     steps:
     - uses: actions/checkout@v4

.github/workflows/backport.yml

@@ -60,7 +60,7 @@ jobs:
           echo "${milestones[@]}"
 
           for i in "${!milestones[@]}"; do
-            if ! git branch -r | grep -q "origin/${milestones[$i]}\$"; then
+            if ! git ls-remote --heads | grep -q "refs/heads/${milestones[$i]}\$"; then
               unset 'milestones[$i]'
             fi
           done

.github/workflows/default.yml

@@ -12,7 +12,7 @@ concurrency:
 
 jobs:
   lint:
-    if: ${{ !contains(github.event.pull_request.labels.*.name, 'skip:ci') && github.event.pull_request.merged == false }}
+    if: ${{ !contains(github.event.pull_request.labels.*.name, 'skip:ci') && !contains(fromJSON('["flow:merge-queue", "flow:hotfix"]'), github.event.label.name) && github.event.pull_request.merged == false }}
     runs-on: ubuntu-latest
     steps:
     - name: Calculate the fetch depth
@@ -79,7 +79,7 @@ jobs:
 
 
   typecheck:
-    if: ${{ !contains(github.event.pull_request.labels.*.name, 'skip:ci') && github.event.pull_request.merged == false }}
+    if: ${{ !contains(github.event.pull_request.labels.*.name, 'skip:ci') && !contains(fromJSON('["flow:merge-queue", "flow:hotfix"]'), github.event.label.name) && github.event.pull_request.merged == false }}
     runs-on: ubuntu-latest
     steps:
     - name: Calculate the fetch depth
@@ -142,7 +142,7 @@ jobs:
 
 
   test:
-    if: ${{ !contains(github.event.pull_request.labels.*.name, 'skip:ci') && github.event.pull_request.merged == false }}
+    if: ${{ !contains(github.event.pull_request.labels.*.name, 'skip:ci') && !contains(fromJSON('["flow:merge-queue", "flow:hotfix"]'), github.event.label.name) && github.event.pull_request.merged == false }}
     runs-on: [ubuntu-latest-8-cores]
     steps:
     - name: Calculate the fetch depth
@@ -295,7 +295,7 @@ jobs:
         name: local-proxy-${{ matrix.os }}
         path: dist-local-proxy/*
     - name: Bootstrap Pants
-      uses: pantsbuild/actions/init-pants@v5-scie-pants
+      uses: pantsbuild/actions/init-pants@v8
       with:
         gha-cache-key: pants-cache-main-1-deploy-py${{ env.PROJECT_PYTHON_VERSION }}-${{ runner.os }}-${{ runner.arch }}
         named-caches-hash: ${{ hashFiles('python*.lock', 'tools/*.lock') }}
@@ -353,7 +353,7 @@ jobs:
       run: |
         pip install -U 'packaging>=21.3'
     - name: Bootstrap Pants
-      uses: pantsbuild/actions/init-pants@v5-scie-pants
+      uses: pantsbuild/actions/init-pants@v8
       with:
         gha-cache-key: pants-cache-main-1-deploy-py${{ env.PROJECT_PYTHON_VERSION }}-${{ runner.os }}-${{ runner.arch }}
         named-caches-hash: ${{ hashFiles('python*.lock', 'tools/*.lock') }}
@@ -401,6 +401,8 @@ jobs:
     needs: [build-scies, build-wheels, build-sbom]
     if: github.event_name == 'push' && contains(github.ref, 'refs/tags/')
     runs-on: ubuntu-latest
+    permissions:
+      contents: write
     environment: deploy-to-pypi
     steps:
     - uses: actions/checkout@v4
@@ -448,7 +450,7 @@ jobs:
         name: SBOM report
         path: dist
     - name: Release to GitHub
-      uses: softprops/action-gh-release@v1
+      uses: softprops/action-gh-release@v2
       with:
         body_path: "CHANGELOG_RELEASE.md"
         prerelease: ${{ env.IS_PRERELEASE }}

.github/workflows/docs-preview.yml

@@ -8,7 +8,7 @@ permissions:
 
 jobs:
   docs-preview-links-:
-    if: ${{ contains(github.event.pull_request.labels.*.name, 'area:docs') && github.event.pull_request.merged == false }}
+    if: ${{ contains(github.event.pull_request.labels.*.name, 'area:docs') && !contains(fromJSON('["flow:merge-queue", "flow:hotfix"]'), github.event.label.name) && github.event.pull_request.merged == false }}
     runs-on: ubuntu-latest
     steps:
       - name: Make a link to the doc preview build (en)

.github/workflows/timeline-check.yml

@@ -4,10 +4,12 @@ on:
   pull_request:
     types: [labeled, unlabeled, opened, synchronize, reopened]
   merge_group:
+permissions:
+  contents: write
 
 jobs:
   pr-number-assign:
-    if: ${{ !contains(github.event.pull_request.labels.*.name, 'skip:changelog') && github.event.pull_request.number != null && github.event.pull_request.merged == false }}
+    if: ${{ !contains(github.event.pull_request.labels.*.name, 'skip:changelog') && !contains(fromJSON('["flow:merge-queue", "flow:hotfix"]'), github.event.label.name) && github.event.pull_request.number != null && github.event.pull_request.merged == false }}
     uses: ./.github/workflows/pr-number-assign.yml
     secrets:
       WORKFLOW_PAT: ${{ secrets.WORKFLOW_PAT }}

.github/workflows/update-api-schema.yml

@@ -8,6 +8,8 @@ on:
 jobs:
   graphql-updated:
     runs-on: ubuntu-latest
+    permissions:
+      contents: write
     steps:
     - name: Calculate the fetch depth
       run: |
@@ -20,13 +22,14 @@ jobs:
       with:
         fetch-depth: ${{ env.GIT_FETCH_DEPTH }}
         ref: ${{ github.head_ref }}
+        token: ${{ secrets.OCTODOG }}
     - name: Extract Python version from pants.toml
       run: |
         PYTHON_VERSION=$(grep -m 1 -oP '(?<=CPython==)([^"]+)' pants.toml)
         echo "PANTS_CONFIG_FILES=pants.ci.toml" >> $GITHUB_ENV
         echo "PROJECT_PYTHON_VERSION=$PYTHON_VERSION" >> $GITHUB_ENV
     - name: Set up Python as Runtime
-      uses: actions/setup-python@v4
+      uses: actions/setup-python@v5
       with:
         python-version: ${{ env.PROJECT_PYTHON_VERSION }}
     - name: Set up remote cache backend (if applicable)
@@ -68,9 +71,13 @@ jobs:
     needs: graphql-updated
     name: Check Schema
     runs-on: ubuntu-latest
+    permissions:
+      contents: read 
+      pull-requests: write
+      checks: write
     steps:
       - uses: actions/checkout@v4
-      - uses: kamilkisiela/graphql-inspector@release-1689086705050
+      - uses: kamilkisiela/graphql-inspector@release-1717403590269
         with:
           schema: 'main:src/ai/backend/manager/api/schema.graphql'
           rules: |

changes/1832.fix.md

@@ -0,0 +1 @@
+Do not omit to update session's occupying resources to DB when a kernel starts.

changes/2041.feature.md

@@ -0,0 +1 @@
+New redis client (experimental)

changes/2128.fix.md

@@ -0,0 +1 @@
+Fix wrong SQL query build for GQL Relay node

changes/2161.fix.md

@@ -0,0 +1 @@
+Fix buggy resolver of `model_card` GQL Query.

changes/2205.fix.md

@@ -0,0 +1 @@
+Ensure that utilization idleness is checked after a set period.

changes/2220.feature.md

@@ -0,0 +1 @@
+Add support for CentOS 8 based kernels

changes/2250.fix.md

@@ -0,0 +1 @@
+Fix GraphQL to support query to non-installed images

changes/2255.feature.md

@@ -0,0 +1 @@
+Allow modifying model service session's environment variable setup

changes/2256.feature.md

@@ -0,0 +1 @@
+Add `endpoint.runtime_variant` column

changes/2258.feature.md

@@ -0,0 +1 @@
+Add new API to show list of supported inference runtimes

changes/2260.feature.md

@@ -0,0 +1 @@
+Add support for model service provisioning without `model-definition.yaml`

changes/2287.fix.md

@@ -0,0 +1 @@
+Rename no-op `access_key` parameter of `endpoint_list` GQL Query to `user_uuid`

changes/2288.fix.md

@@ -0,0 +1 @@
+Fix `ai.backend.service-ports` label syntax broken when image does not expose built-in service port

changes/2289.fix.md

@@ -0,0 +1 @@
+Improve stability of `untag_image_from_registry` mutation

changes/2290.fix.md

@@ -0,0 +1 @@
+SSH not working between kernels started with customized image

changes/2291.fix.md

@@ -0,0 +1 @@
+Invalid container memory capacity reported

changes/2318.fix.md

@@ -0,0 +1 @@
+Fix model service sessions created before 24.03.5 failing to spawn

changes/2319.fix.md

@@ -0,0 +1 @@
+Image commit not working

changes/2320.fix.md

@@ -0,0 +1 @@
+model service session scheduler (`scale_services()`) failing when sessions bound to active route already marked as terminated

changes/2321.fix.md

@@ -0,0 +1 @@
+Fix container metric collection halted on systems with Cgroups v1

docs/release-notes/24.03.md

@@ -0,0 +1,29 @@
+# What's New
+
+## Neo WebUI
+
+This version introduces a next-generation WebUI preview with the dark-mode support.
+You may toggle the switch in the session list view to use the "neo" session list and launcher,
+which provides more streamlined UI.
+
+
+## TUI Installer
+
+From this release, we ship an open-source version TUI installer as a self-contained single-binary executable.
+This installer is packaged using [the science project](https://github.com/a-scie) and provides a terminal-based GUI (aka TUI) for easier setup.
+
+
+## Model Store
+
+It introduces a globally shared predefined "model-store" project where users may clone the model vfolders to their accounts.
+
+
+## VFolder Trash Bin
+
+When users delete a vfolder, it is now sent to the trash bin instead of immediately removing all its contents. 
+This allows users or admins to undo accidental deletions, and decouples the storage backend's directory removal process from the vfolder management system.
+
+
+## User-defined Image Commit
+
+When allowed by the administrator, users may commit a running session's main container as a new container image.

docs/release-notes/24.09.md

@@ -0,0 +1,47 @@
+# What's New
+
+## Model Serving
+
+(TODO)
+<!-- auto-scaling -->
+
+
+## Model Store
+
+This release upgrades the model store with the metadata browser and improved UI.
+Users can now import and run a model directly from configured external model repository services like HuggingFace.
+
+
+## NVIDIA NIM Integration
+
+If users have a valid license for NVIDIA NIM, they can launch a NIM container with one click as a model service.
+
+
+## Fine-grained Access Control
+
+(TODO)
+<!--
+Now projects may have their own administrators who can add or remove users belonging to the project and project-owned vfolders without manual intervention by the super-administrator.
+Under the hood, the storage management subsystem has adopted a fine-grained RBAC to declare and query the privilege for a user to perform a specific action on a target object,
+allowing more fine-grained customization of storage access policies for large enterprises.
+-->
+
+
+## FastTrack
+
+(TODO)
+<!-- project-wide collaboration -->
+
+
+## Neo WebUI
+
+This version continues transition to the next-generation WebUI.
+(TODO: more details)
+
+
+<!--
+## Cancellable Image Pulling
+
+Users can now trigger and cancel the session creation process and the required image pulling process separately,
+helping the users to manage sessions stuck in the "PULLING" status due to various reasons (NFS mount failures, a broken Docker daemon, etc.).
+-->

pants.toml

@@ -65,6 +65,7 @@ use_rust_parser = true
 
 [python-repos]
 indexes = ["https://dist.backend.ai/pypi/simple/", "https://pypi.org/simple/"]
+find_links = ["file://%(buildroot)s/wheelhouse"]
 
 [python.resolves]
 python-default = "python.lock"

scripts/agent/build-dropbear.sh

@@ -2,7 +2,7 @@
 set -e
 
 arch=$(uname -m)
-distros=("ubuntu18.04" "ubuntu20.04" "ubuntu22.04" "alpine3.8")
+distros=("alpine3.8" "centos8.0" "ubuntu18.04" "ubuntu20.04" "ubuntu22.04")
 
 ubuntu1804_builder_dockerfile=$(cat <<'EOF'
 FROM ubuntu:18.04
@@ -31,6 +31,16 @@ RUN apk add --no-cache make gcc musl-dev
 RUN apk add --no-cache autoconf automake zlib-dev
 EOF
 )
+centos8_builder_dockerfile=$(cat <<'EOF'
+FROM centos:centos8
+RUN sed -i 's|#baseurl=http://mirror.centos.org|baseurl=http://vault.centos.org|g' /etc/yum.repos.d/CentOS-Linux-*
+
+RUN dnf install -y make gcc automake autoconf dnf-plugins-core
+RUN dnf config-manager --set-enabled powertools
+RUN dnf install -y zlib-static glibc-static libxcrypt-static
+EOF
+)
+
 
 build_script=$(cat <<'EOF'
 #! /bin/sh
@@ -67,6 +77,7 @@ echo "$ubuntu1804_builder_dockerfile" > "$SCRIPT_DIR/dropbear-builder.ubuntu18.0
 echo "$ubuntu2004_builder_dockerfile" > "$SCRIPT_DIR/dropbear-builder.ubuntu20.04.dockerfile"
 echo "$ubuntu2204_builder_dockerfile" > "$SCRIPT_DIR/dropbear-builder.ubuntu22.04.dockerfile"
 echo "$alpine_builder_dockerfile" > "$SCRIPT_DIR/dropbear-builder.alpine3.8.dockerfile"
+echo "$centos8_builder_dockerfile" > "$SCRIPT_DIR/dropbear-builder.centos8.0.dockerfile"
 
 for distro in "${distros[@]}"; do
   docker build -t dropbear-builder:$distro \

scripts/agent/build-sftpserver.sh

@@ -2,10 +2,10 @@
 set -e
 
 arch=$(uname -m)
-distros=("ubuntu16.04" "ubuntu18.04" "ubuntu20.04" "centos7.6" "alpine3.8")
+distros=("alpine3.8" "centos7.6" "centos8.0" "ubuntu16.04" "ubuntu18.04" "ubuntu20.04")
 
 static_libs_dockerfile_part=$(cat <<'EOF'
-ENV ZLIB_VER=1.2.11 \
+ENV ZLIB_VER=1.3.1 \
     SSL_VER=1.1.1i
 
 RUN wget https://www.zlib.net/zlib-${ZLIB_VER}.tar.gz -O /root/zlib-${ZLIB_VER}.tar.gz && \
@@ -70,6 +70,16 @@ RUN yum install -y wget
 EOF
 )
 
+centos8_builder_dockerfile=$(cat <<'EOF'
+FROM centos:centos8
+RUN sed -i 's|#baseurl=http://mirror.centos.org|baseurl=http://vault.centos.org|g' /etc/yum.repos.d/CentOS-Linux-*
+
+RUN dnf install -y make gcc
+RUN dnf install -y autoconf
+RUN dnf install -y wget
+EOF
+)
+
 alpine_builder_dockerfile=$(cat <<'EOF'
 FROM alpine:3.8
 RUN apk add --no-cache make gcc musl-dev
@@ -107,6 +117,7 @@ echo -e "$ubuntu1604_builder_dockerfile\n$static_libs_dockerfile_part" > "$SCRIP
 echo -e "$ubuntu1804_builder_dockerfile\n$static_libs_dockerfile_part" > "$SCRIPT_DIR/sftpserver-builder.ubuntu18.04.dockerfile"
 echo -e "$ubuntu2004_builder_dockerfile\n$static_libs_dockerfile_part" > "$SCRIPT_DIR/sftpserver-builder.ubuntu20.04.dockerfile"
 echo -e "$centos_builder_dockerfile\n$static_libs_dockerfile_part" > "$SCRIPT_DIR/sftpserver-builder.centos7.6.dockerfile"
+echo -e "$centos8_builder_dockerfile\n$static_libs_dockerfile_part" > "$SCRIPT_DIR/sftpserver-builder.centos8.0.dockerfile"
 echo -e "$alpine_builder_dockerfile\n$static_libs_dockerfile_part" > "$SCRIPT_DIR/sftpserver-builder.alpine3.8.dockerfile"
 
 for distro in "${distros[@]}"; do
@@ -129,8 +140,8 @@ for distro in "${distros[@]}"; do
 done
 
 ls -l .
-cp sftp-server.*.bin $SCRIPT_DIR/../src/ai/backend/runner
-cp scp.*.bin $SCRIPT_DIR/../src/ai/backend/runner
+cp sftp-server.*.bin $SCRIPT_DIR/../../src/ai/backend/runner
+cp scp.*.bin $SCRIPT_DIR/../../src/ai/backend/runner
 
 cd $SCRIPT_DIR/..
 rm -rf "$temp_dir"

scripts/agent/build-suexec.sh

@@ -2,8 +2,8 @@
 set -e
 
 arch=$(uname -m)
-# distros=("ubuntu16.04" "ubuntu18.04" "ubuntu20.04" "centos7.6" "alpine3.8")
-distros=("ubuntu16.04")
+
+distros=("alpine3.8" "centos7.6" "centos8.0" "ubuntu16.04" "ubuntu18.04" "ubuntu20.04")
 
 if [ $arch = "arm64" ]; then
   arch="aarch64"
@@ -36,6 +36,14 @@ RUN yum install -y make gcc
 EOF
 )
 
+centos8_builder_dockerfile=$(cat <<'EOF'
+FROM centos:centos8
+RUN sed -i 's|#baseurl=http://mirror.centos.org|baseurl=http://vault.centos.org|g' /etc/yum.repos.d/CentOS-Linux-*
+
+RUN dnf install -y make gcc
+EOF
+)
+
 alpine_builder_dockerfile=$(cat <<'EOF'
 FROM alpine:3.8
 RUN apk add --no-cache make gcc musl-dev
@@ -61,6 +69,7 @@ echo "$ubuntu1604_builder_dockerfile" > "$SCRIPT_DIR/suexec-builder.ubuntu16.04.
 echo "$ubuntu1804_builder_dockerfile" > "$SCRIPT_DIR/suexec-builder.ubuntu18.04.dockerfile"
 echo "$ubuntu2004_builder_dockerfile" > "$SCRIPT_DIR/suexec-builder.ubuntu20.04.dockerfile"
 echo "$centos_builder_dockerfile" > "$SCRIPT_DIR/suexec-builder.centos7.6.dockerfile"
+echo "$centos8_builder_dockerfile" > "$SCRIPT_DIR/suexec-builder.centos8.0.dockerfile"
 echo "$alpine_builder_dockerfile" > "$SCRIPT_DIR/suexec-builder.alpine3.8.dockerfile"
 
 for distro in "${distros[@]}"; do