build-container.yml

---
# yamllint disable rule:line-length
name: "CI"
on:  # yamllint disable rule:truthy
  push:
    tags:
      - '*'
    branches:
      - '*'
  pull_request:
    branches:
      - devel
  schedule:
    - cron: '0 0 * * *'
  workflow_dispatch:
  release:
    types:
      - created
# Credit: https://stackoverflow.com/a/72408109
concurrency:
  group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
  cancel-in-progress: true
env:
  WAIT_INTERVAL_SECS: 1
jobs:
  build:
    runs-on: ubuntu-latest
    steps:
      - name: Checkout code
        uses: actions/checkout@v4
        with:
          submodules: recursive

      - name: Build container images - spdk
        run: make build SVC="spdk" SPDK_TARGET_ARCH=x86-64-v2

      - name: Build container images - bdevperf
        run: make build SVC="bdevperf" SPDK_TARGET_ARCH=x86-64-v2

      - name: Build container images - nvmeof
        run: make build SVC="nvmeof" SPDK_TARGET_ARCH=x86-64-v2

      - name: Build container images - nvmeof-cli
        run: make build SVC="nvmeof-cli" SPDK_TARGET_ARCH=x86-64-v2

      - name: Save container images
        run: |
          . .env
          docker save $QUAY_NVMEOF:$NVMEOF_VERSION > nvmeof.tar
          docker save $QUAY_NVMEOFCLI:$NVMEOF_VERSION > nvmeof-cli.tar
          docker save bdevperf > bdevperf.tar

      - name: Upload container images
        uses: actions/upload-artifact@v4
        with:
          name: container_images
          path: |
            nvmeof.tar
            nvmeof-cli.tar
            bdevperf.tar

  build-ceph:
    runs-on: ubuntu-latest
    steps:
      - name: Checkout code
        uses: actions/checkout@v4

      - name: Build container images - ceph
        run: make build SVC=ceph

      - name: Save container images
        run: |
          . .env
          docker save $QUAY_CEPH:$CEPH_VERSION > ceph.tar

      - name: Upload container images
        uses: actions/upload-artifact@v4
        with:
          name: container_images_ceph
          path: |
            ceph.tar

  pytest:
    needs: [build, build-ceph]
    strategy:
      fail-fast: false
      matrix:
        test: ["cli", "cli_change_lb", "state", "multi_gateway", "server", "grpc", "omap_lock", "old_omap", "log_files", "nsid"]
    runs-on: ubuntu-latest
    env:
      HUGEPAGES: 512  # for multi gateway test, approx 256 per gateway instance
    steps:
      - name: Checkout code
        uses: actions/checkout@v4

      - name: Setup huge pages
        run: |
          make setup HUGEPAGES=$HUGEPAGES

      - name: Download container images
        uses: actions/download-artifact@v4
        with:
          pattern: container_images*
          merge-multiple: true

      - name: Load container images
        run: |
          docker load < nvmeof.tar
          docker load < ceph.tar

      - name: Clear space on disk
        run: |
          if [[ -d /usr/share/dotnet ]]; then
            /usr/bin/du -sh /usr/share/dotnet
            rm -rf /usr/share/dotnet
          fi
          if [[ -d /opt/ghc ]]; then
            /usr/bin/du -sh /opt/ghc
            rm -rf /opt/ghc
          fi
          if [[ -d /usr/local/share/boost ]]; then
            /usr/bin/du -sh /usr/local/share/boost
            rm -rf /usr/local/share/boost
          fi
          if [[ -n "$AGENT_TOOLSDIRECTORY" ]]; then
            if [[ -d "$AGENT_TOOLSDIRECTORY" ]]; then
              /usr/bin/du -sh "$AGENT_TOOLSDIRECTORY"
              rm -rf "$AGENT_TOOLSDIRECTORY"
            fi
          fi

      - name: Start ceph cluster
        run: |
          if ! docker-compose --version 2>&1 > /dev/null ; then
            sudo apt update
            sudo apt install -y docker-compose
          fi
          docker-compose --version
          docker-compose up -d ceph

      - name: Wait for the ceph cluster container to become healthy
        timeout-minutes: 3
        run: |
          while true; do
            container_status=$(docker inspect --format='{{.State.Health.Status}}' ceph)
            if [[ $container_status == "healthy" ]]; then
              # success
              exit 0
            else
              # Wait for a specific time before checking again
              sleep ${{ env.WAIT_INTERVAL_SECS }}
              echo -n .
            fi
          done

      - name: Create RBD image
        run: |
          echo "💁 ceph create pools:"
          make exec SVC=ceph OPTS="-T" CMD="ceph osd pool create rbd"
          make exec SVC=ceph OPTS="-T" CMD="ceph osd pool create rbd2"
          echo "💁 ceph list pools:"
          make exec SVC=ceph OPTS="-T" CMD="ceph osd lspools"
          echo "💁 rbd create:"
          make exec SVC=ceph OPTS="-T" CMD="rbd create rbd/mytestdevimage --size 16"
          make exec SVC=ceph OPTS="-T" CMD="rbd create rbd2/mytestdevimage2 --size 16"
          echo "💁 ls rbd:"
          make exec SVC=ceph OPTS="-T" CMD="rbd ls rbd"
          make exec SVC=ceph OPTS="-T" CMD="rbd ls rbd2"

      - name: Run protoc
        run: |
          make protoc

      - name: Run ${{ matrix.test }} test
        run: |
          # Run tests code in current dir
          # Managing pytest’s output: https://docs.pytest.org/en/7.1.x/how-to/output.html
          make run SVC="nvmeof" OPTS="--volume=$(pwd)/tests:/src/tests --entrypoint=python3" CMD="-m pytest --show-capture=all -s --full-trace -vv -rA tests/test_${{ matrix.test }}.py"

      - name: Check coredump existence
        if: success() || failure()
        id: check_coredumps
        uses: andstor/file-existence-action@20b4d2e596410855db8f9ca21e96fbe18e12930b  # v2, pinned to SHA for security reasons
        with:
          files: "/tmp/coredump/core.*"

      - name: Upload ${{ matrix.test }} test core dumps
        if: steps.check_coredumps.outputs.files_exists == 'true'
        uses: actions/upload-artifact@v4
        with:
          name: core_pytest_${{ matrix.test }}
          path: /tmp/coredump/core.*

      - name: Display logs
        if: success() || failure()
        run: |
          make logs OPTS=""

      - name: Tear down
        if: success() || failure()
        run: |
          make down
          make clean

  demo:
    needs: [build, build-ceph]
    runs-on: ubuntu-latest
    env:
      HUGEPAGES: 512
    steps:
      - name: Checkout code
        uses: actions/checkout@v4

      - name: Setup huge-pages
        run: make setup HUGEPAGES=$HUGEPAGES

      - name: Download container images
        uses: actions/download-artifact@v4
        with:
          pattern: container_images*
          merge-multiple: true

      - name: Load container images
        run: |
          docker load < nvmeof.tar
          docker load < nvmeof-cli.tar
          docker load < ceph.tar
          docker load < bdevperf.tar

      - name: Start containers
        timeout-minutes: 3
        run: |
          if ! docker-compose --version 2>&1 > /dev/null ; then
            sudo apt update
            sudo apt install -y docker-compose
          fi
          docker-compose --version
          make up

      - name: Wait for the Gateway to be listening
        timeout-minutes: 3
        run: |
          . .env

          echo using gateway $NVMEOF_IP_ADDRESS port $NVMEOF_GW_PORT
          until nc -z $NVMEOF_IP_ADDRESS $NVMEOF_GW_PORT; do
            echo -n .
            sleep ${{ env.WAIT_INTERVAL_SECS }}
          done

      - name: List containers
        if: success() || failure()
        run: make ps

      - name: List processes
        if: success() || failure()
        run: make top

      - name: Test
        run: |
          make demo OPTS=-T NVMEOF_CONTAINER_NAME="ceph-nvmeof_nvmeof_1"

      - name: List resources
        run: |
          # https://github.com/actions/toolkit/issues/766
          shopt -s expand_aliases
          eval $(make alias)
          cephnvmf subsystem list
          subs=$(cephnvmf --output stdio --format json subsystem list | grep nqn | sed 's/"nqn": "//' | sed 's/",$//')
          for sub in $subs
          do
              cephnvmf namespace list --subsystem $sub
              cephnvmf listener list --subsystem $sub
              cephnvmf host list --subsystem $sub
          done

      - name: Run bdevperf
        run: |
          # see https://spdk.io/doc/nvmf_multipath_howto.html
          shopt -s expand_aliases
          eval $(make alias)
          . .env
          set -x
          echo -n "ℹ️  Starting bdevperf container"
          docker-compose up -d bdevperf
          sleep 10
          echo "ℹ️  bdevperf start up logs"
          make logs SVC=bdevperf
          eval $(make run SVC=bdevperf OPTS="--entrypoint=env" | grep BDEVPERF_SOCKET | tr -d '\n\r' )

          rpc="/usr/libexec/spdk/scripts/rpc.py"
          echo "ℹ️  bdevperf bdev_nvme_set_options"
          make exec SVC=bdevperf OPTS=-T CMD="$rpc -v -s $BDEVPERF_SOCKET bdev_nvme_set_options -r -1"
          echo "ℹ️  bdevperf tcp connect ip: $NVMEOF_IP_ADDRESS port: $NVMEOF_IO_PORT nqn: $NQN"
          make exec SVC=bdevperf OPTS=-T CMD="$rpc -v -s $BDEVPERF_SOCKET bdev_nvme_attach_controller -b Nvme0 -t tcp -a $NVMEOF_IP_ADDRESS -s $NVMEOF_IO_PORT -f ipv4 -n $NQN -q ${NQN}host -l -1 -o 10"
          echo "ℹ️  verify connection list"
          conns=$(cephnvmf --output stdio --format json connection list --subsystem $NQN)
          echo $conns | grep -q '"status": 0'
          echo $conns | grep -q "\"nqn\": \"${NQN}host\""
          echo $conns | grep -q "\"trsvcid\": ${NVMEOF_IO_PORT}"
          echo $conns | grep -q "\"traddr\": \"${NVMEOF_IP_ADDRESS}\""
          echo $conns | grep -q "\"adrfam\": \"ipv4\""
          echo $conns | grep -q "\"trtype\": \"TCP\""
          con_cnt=$(echo $conns | xargs -n 1 | grep traddr | wc -l)
          if [ $con_cnt -ne 1 ]; then
              echo "Number of connections ${con_cnt}, expected 1 list: ${conns}"
              exit 1
          fi
          echo $conns | grep -q "\"qpairs_count\": 1"
          echo $conns | grep -q "\"connected\": true"
          echo "ℹ️  bdevperf perform_tests"
          eval $(make run SVC=bdevperf OPTS="--entrypoint=env" | grep BDEVPERF_TEST_DURATION | tr -d '\n\r' )
          timeout=$(expr $BDEVPERF_TEST_DURATION \* 2)
          bdevperf="/usr/libexec/spdk/scripts/bdevperf.py"
          make exec SVC=bdevperf OPTS=-T CMD="$bdevperf -v -t $timeout -s $BDEVPERF_SOCKET perform_tests"

      - name: Check coredump existence
        if: success() || failure()
        id: check_coredumps
        uses: andstor/file-existence-action@20b4d2e596410855db8f9ca21e96fbe18e12930b  # v2, pinned to SHA for security reasons
        with:
          files: "/tmp/coredump/core.*"

      - name: Upload demo core dumps
        if: steps.check_coredumps.outputs.files_exists == 'true'
        uses: actions/upload-artifact@v4
        with:
          name: core_demo
          path: /tmp/coredump/core.*

      # For debugging purposes (provides an SSH connection to the runner)
      # - name: Setup tmate session
      #   uses: mxschmitt/action-tmate@v3
      #   with:
      #     limit-access-to-actor: true

      - name: Display logs
        if: success() || failure()
        run: make logs OPTS=''

      - name: Tear down
        if: success() || failure()
        run: |
          make down
          make clean

  discovery:
    needs: [build, build-ceph]
    strategy:
      fail-fast: false
      matrix:
        integration: ["container", "embedded"]
    runs-on: ubuntu-latest
    env:
      HUGEPAGES: 768  # 3 spdk instances

    steps:
      - name: Checkout code
        uses: actions/checkout@v4

      - name: Setup huge-pages
        run: make setup HUGEPAGES=$HUGEPAGES

      - name: Download container images
        uses: actions/download-artifact@v4
        with:
          pattern: container_images*
          merge-multiple: true

      - name: Load container images
        run: |
          docker load < nvmeof.tar
          docker load < nvmeof-cli.tar
          docker load < ceph.tar
          docker load < bdevperf.tar

      - name: Start discovery controller
        if: matrix.integration == 'container'
        run: |
          if ! docker-compose --version 2>&1 > /dev/null ; then
            sudo apt update
            sudo apt install -y docker-compose
          fi
          docker-compose --version
          docker-compose up --detach discovery

      - name: Wait for discovery controller to be listening
        if: matrix.integration == 'container'
        timeout-minutes: 3
        run: |
          . .env
          container_ip() {
            docker inspect -f '{{range .NetworkSettings.Networks}}{{.IPAddress}}{{end}}' "$1"
          }

          ip=$(container_ip $DISC1)
          echo using discovery controller $ip $NVMEOF_DISC_PORT
          until nc -z $ip $NVMEOF_DISC_PORT; do
            echo -n .
            sleep ${{ env.WAIT_INTERVAL_SECS }}
          done

      - name: Start gateway with scale=2
        run: |
          ./tests/ha/start_up.sh

      - name: Wait for gateways to be listening
        timeout-minutes: 3
        run: |
          . .env
          container_ip() {
            docker inspect -f '{{range .NetworkSettings.Networks}}{{.IPAddress}}{{end}}' "$1"
          }

          for gw in $GW1 $GW2; do
            ip=$(container_ip $gw)
            echo using gateway $ip $NVMEOF_GW_PORT
            until nc -z $ip $NVMEOF_GW_PORT; do
              echo -n .
              sleep ${{ env.WAIT_INTERVAL_SECS }}
            done
            echo
          done

      - name: List containers
        if: success() || failure()
        run: |
          docker-compose ps

      - name: List processes
        if: success() || failure()
        run: |
          docker-compose top

      - name: Set up target
        run: |
          . .env

          container_ip() {
            docker inspect -f '{{range .NetworkSettings.Networks}}{{.IPAddress}}{{end}}' "$1"
          }
          # container id is the default hostname in docker environent
          # i.e. default gateway name
          container_id() {
            docker ps -q -f name=$1
          }
          cli_gw() {
            gw=$1
            shift
            docker-compose run --rm nvmeof-cli --server-address $gw --server-port $NVMEOF_GW_PORT "$@"
          }

          gw1=$(container_ip $GW1)
          echo ℹ️ Using GW RPC $GW1 address $gw1 port $NVMEOF_GW_PORT
          cli_gw $gw1 subsystem list
          cli_gw $gw1 subsystem add --subsystem $NQN --serial $SERIAL
          cli_gw $gw1 namespace add --subsystem $NQN --rbd-pool $RBD_POOL --rbd-image $RBD_IMAGE_NAME --size $RBD_IMAGE_SIZE --rbd-create-image
          for gw in $GW1 $GW2; do
            ip=$(container_ip $gw)
            name=$(container_id $gw) # default hostname - container id
            echo ℹ️ Create listener address $ip gateway $name
            cli_gw $ip listener add --subsystem $NQN --host-name $name --traddr $ip --trsvcid $NVMEOF_IO_PORT
          done
          cli_gw $gw1 host add --subsystem $NQN --host "*"
          for gw in $GW1 $GW2; do
            ip=$(container_ip $gw)
            echo ℹ️ Subsystems for name $gw ip $ip
            cli_gw $ip subsystem list
          done

      - name: Run bdevperf discovery
        run: |
          # See
          # - https://github.com/spdk/spdk/blob/master/doc/jsonrpc.md
          # - https://spdk.io/doc/nvmf_multipath_howto.html
          . .env
          container_ip() {
            docker inspect -f '{{range .NetworkSettings.Networks}}{{.IPAddress}}{{end}}' "$1"
          }

          echo -n "ℹ️  Starting bdevperf container"
          docker-compose up -d bdevperf
          sleep 10
          echo "ℹ️  bdevperf start up logs"
          make logs SVC=bdevperf
          eval $(make run SVC=bdevperf OPTS="--entrypoint=env" | grep BDEVPERF_SOCKET | tr -d '\n\r' )

          if [ "${{ matrix.integration }}" == "embedded" ]; then
            ip=$(container_ip $GW1)
            echo "ℹ️  Using discovery service in gateway $GW1 ip  $ip"
          else
            ip=$(container_ip $DISC1)
            echo "ℹ️  Using standalone discovery container $DISC1 ip  $ip"
          fi
          rpc="/usr/libexec/spdk/scripts/rpc.py"
          echo "ℹ️  bdevperf bdev_nvme_set_options"
          make exec SVC=bdevperf OPTS=-T CMD="$rpc -v -s $BDEVPERF_SOCKET bdev_nvme_set_options -r -1"
          echo "ℹ️  bdevperf start discovery ip: $ip port: $NVMEOF_DISC_PORT"
          # -l -1 -o 10
          make exec SVC=bdevperf OPTS=-T CMD="$rpc -v -s $BDEVPERF_SOCKET bdev_nvme_start_discovery -b Nvme0 -t tcp -a $ip -s $NVMEOF_DISC_PORT -f ipv4 -w"
          echo "ℹ️  bdevperf bdev_nvme_get_discovery_info"
          make exec SVC=bdevperf OPTS=-T CMD="$rpc -v -s $BDEVPERF_SOCKET bdev_nvme_get_discovery_info"
          echo "ℹ️  bdevperf perform_tests"
          eval $(make run SVC=bdevperf OPTS="--entrypoint=env" | grep BDEVPERF_TEST_DURATION | tr -d '\n\r' )
          timeout=$(expr $BDEVPERF_TEST_DURATION \* 2)
          bdevperf="/usr/libexec/spdk/scripts/bdevperf.py"
          make exec SVC=bdevperf OPTS=-T CMD="$bdevperf -v -t $timeout -s $BDEVPERF_SOCKET perform_tests"

      - name: Check coredump existence
        if: success() || failure()
        id: check_coredumps
        uses: andstor/file-existence-action@20b4d2e596410855db8f9ca21e96fbe18e12930b  # v2, pinned to SHA for security reasons
        with:
          files: "/tmp/coredump/core.*"

      - name: Upload demo core dumps
        if: steps.check_coredumps.outputs.files_exists == 'true'
        uses: actions/upload-artifact@v4
        with:
          name: core_demo_discovery
          path: /tmp/coredump/core.*

      - name: Display logs
        if: success() || failure()
        run: make logs OPTS=''

      - name: Tear down
        if: success() || failure()
        run: |
          make down
          make clean

  ha:
    needs: [build, build-ceph]
    strategy:
      fail-fast: false
      matrix:
        test: ["sanity", "state_transitions", "state_transitions_both_gws", "state_transitions_loop", "state_transitions_rand_loop", "late_registration", "late_registration_loop", "4gws", "4gws_loop", "4gws_create_delete", "4gws_create_delete_loop", "namespaces", "namespaces_loop", "mtls", "notify"]
    runs-on: ubuntu-latest
    env:
      HUGEPAGES: 1024  # 4 spdk instances

    steps:
      - name: Checkout code
        uses: actions/checkout@v4

      - name: Setup huge-pages
        run: make setup HUGEPAGES=$HUGEPAGES

      - name: Download container images
        uses: actions/download-artifact@v4
        with:
          pattern: container_images*
          merge-multiple: true

      - name: Load container images
        run: |
          docker load < nvmeof.tar
          docker load < nvmeof-cli.tar
          docker load < ceph.tar
          docker load < bdevperf.tar

      - name: Start HA with two gateways
        timeout-minutes: 3
        run: |
          if ! docker-compose --version 2>&1 > /dev/null ; then
            sudo apt update
            sudo apt install -y docker-compose
          fi
          docker-compose --version
          test_specific_start_up="./tests/ha/start_up_${{ matrix.test }}.sh"
          if [ -x "$test_specific_start_up" ] || [ -L "$test_specific_start_up" ] && [ -x "$(readlink -f "$test_specific_start_up")" ]; then
            $test_specific_start_up
          else
            ./tests/ha/start_up.sh
          fi

      - name: Wait for gateways to be listening
        timeout-minutes: 3
        run: |
          test_specific_wait="./tests/ha/wait_gateways_${{ matrix.test }}.sh"
          if [ -x "$test_specific_wait" ] || [ -L "$test_specific_wait" ] && [ -x "$(readlink -f "$test_specific_wait")" ]; then
            $test_specific_wait
          else
            ./tests/ha/wait_gateways.sh
          fi

      - name: List containers
        if: success() || failure()
        run: |
          docker-compose ps

      - name: List processes
        if: success() || failure()
        run: |
          docker-compose top

      - name: Set up target
        run: |
          test_specific_setup="./tests/ha/setup_${{ matrix.test }}.sh"
          if [ -x "$test_specific_setup" ] || [ -L "$test_specific_setup" ] && [ -x "$(readlink -f "$test_specific_setup")" ]; then
            $test_specific_setup
          else
            ./tests/ha/setup.sh
          fi

      - name: Run HA ${{ matrix.test }} test
        timeout-minutes: 30
        run: |
          . .env
          source "tests/ha/${{ matrix.test }}.sh"

      - name: Check coredump existence
        if: success() || failure()
        id: check_coredumps
        uses: andstor/file-existence-action@20b4d2e596410855db8f9ca21e96fbe18e12930b  # v2, pinned to SHA for security reasons
        with:
          files: "/tmp/coredump/core.*"

      - name: Upload ha core dumps
        if: steps.check_coredumps.outputs.files_exists == 'true'
        uses: actions/upload-artifact@v4
        with:
          name: core_demo_ha-${{ matrix.test }}
          path: /tmp/coredump/core.*

      - name: Copy ceph logs
        if: success() || failure()
        run: docker cp  ceph:/ceph/out /tmp/out

      - name: Upload ceph logs
        if: success() || failure()
        uses: actions/upload-artifact@v4
        with:
          name: ceph_out_${{ matrix.test }}
          path: /tmp/out/*

      - name: Display logs
        if: success() || failure()
        run: make logs OPTS=''

      - name: Tear down
        if: success() || failure()
        run: |
          make down
          make clean

  push-images-to-private-registry:
    if: github.event_name == 'push' && github.ref == 'refs/heads/devel' && github.repository == 'ceph/ceph-nvmeof'
    needs: [pytest, demo, discovery, ha]
    runs-on: ubuntu-latest

    steps:
      - name: Checkout code
        uses: actions/checkout@v4

      - name: Download container images
        uses: actions/download-artifact@v4
        with:
          name: container_images

      - name: Load container images
        run: |
          docker load < nvmeof.tar
          docker load < nvmeof-cli.tar

      - name: Login to quay.io
        uses: docker/login-action@v2
        with:
          registry: ${{ vars.CONTAINER_REGISTRY }}
          username: '${{ vars.CONTAINER_PRIVATE_REGISTRY_USERNAME }}'
          password: '${{ secrets.CONTAINER_PRIVATE_REGISTRY_PASSWORD }}'

      - name: Publish nvmeof containers when release/tag is created
        run: |
          make push QUAY="${{ vars.CONTAINER_REGISTRY }}/${{ vars.CONTAINER_PRIVATE_REGISTRY_USERNAME }}"

  push-images-to-ceph-registry:
    if: github.event_name == 'release'
    needs: [pytest, demo, discovery, ha]
    runs-on: ubuntu-latest

    steps:
      - name: Checkout code
        uses: actions/checkout@v4

      - name: Download container images
        uses: actions/download-artifact@v4
        with:
          name: container_images

      - name: Load container images
        run: |
          docker load < nvmeof.tar
          docker load < nvmeof-cli.tar

      - name: Login to quay.io
        uses: docker/login-action@v2
        with:
          registry: ${{ vars.CONTAINER_REGISTRY }}
          username: '${{ vars.CONTAINER_REGISTRY_USERNAME }}'
          password: '${{ secrets.CONTAINER_REGISTRY_PASSWORD }}'

      - name: Publish nvmeof containers when release/tag is created
        run: |
          make push QUAY="${{ vars.CONTAINER_REGISTRY_USERNAME }}"