sanity-test.yaml

name: Sanity Testing
on:
  push:
    branches:
      - main
  workflow_dispatch:
    inputs:
      user:
        description: "User to run sanity test on"

permissions:
  # required to retrieve AWS credentials
  id-token: write
  contents: write

# cancel currently running jobs if a new version of the branch is pushed
concurrency:
  group: sanity_testing-${{ github.workflow }}-${{ github.ref }}
  cancel-in-progress: true

jobs:
  SetM365App:
    uses: alcionai/corso/.github/workflows/accSelector.yaml@main

  Sanity-Tests:
    needs: [SetM365App]
    environment: Testing
    runs-on: ubuntu-latest
    env:
      # Need these in the local env so that corso can read them
      AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
      AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_ACCESS_KEY_SECRET }}
      AZURE_CLIENT_ID: ${{ secrets[needs.SetM365App.outputs.client_id_env] }}
      AZURE_CLIENT_SECRET: ${{ secrets[needs.SetM365App.outputs.client_secret_env] }}
      AZURE_TENANT_ID: ${{ secrets.TENANT_ID }}
      CORSO_PASSPHRASE: ${{ secrets.INTEGRATION_TEST_CORSO_PASSPHRASE }}
      # re-used values
      CORSO_LOG_DIR: ${{ github.workspace }}/src/testlog
      CORSO_LOG_FILE: ${{ github.workspace }}/src/testlog/run-sanity.log
      RESTORE_DEST_PFX: Corso_Test_Sanity_
      TEST_USER: ${{ github.event.inputs.user != '' && github.event.inputs.user || vars.CORSO_M365_TEST_USER_ID }}

    defaults:
      run:
        working-directory: src

    ##########################################################################################################################################

    # setup
    steps:
      - uses: actions/checkout@v4

      - name: Setup Golang with cache
        uses: magnetikonline/action-golang-cache@v4
        with:
          go-version-file: src/go.mod

      - run: go build -o corso
        timeout-minutes: 10

      - run: go build -o sanity-test ./cmd/sanity_test
        timeout-minutes: 10

      - run: mkdir ${CORSO_LOG_DIR}

      ##########################################################################################################################################

      # Pre-Run cleanup

      # unlike CI tests, sanity tests are not expected to run concurrently.
      # however, the sanity yaml concurrency is set to a maximum of 1 run, preferring
      # the latest release.  If we wait to clean up the production til after the tests
      # It would be possible to complete all the testing but cancel the run before
      # cleanup occurs.  Setting the cleanup before the tests ensures we always begin
      # with a clean slate, and cannot compound data production.
      - name: Set purge boundary
        if: always()
        run: |
          echo "NOW=$(date +"%Y-%m-%dT%H:%M:%SZ")" >> $GITHUB_ENV

      - name: Purge CI-Produced Folders for Users
        timeout-minutes: 30
        uses: ./.github/actions/purge-m365-data
        with:
          user: ${{ env.TEST_USER }}
          folder-prefix: ${{ env.RESTORE_DEST_PFX }}
          older-than: ${{ env.NOW }}
          azure-client-id: ${{ secrets[needs.SetM365App.outputs.client_id_env] }}
          azure-client-secret: ${{ secrets[needs.SetM365App.outputs.client_secret_env] }}
          azure-tenant-id: ${{ secrets.TENANT_ID }}
          m365-admin-user: ${{ secrets.M365_TENANT_ADMIN_USER }}
          m365-admin-password: ${{ secrets.M365_TENANT_ADMIN_PASSWORD }}
          azure-pnp-client-id: ${{ secrets.AZURE_PNP_CLIENT_ID }}
          azure-pnp-client-cert: ${{ secrets.AZURE_PNP_CLIENT_CERT }}
          tenant-domain: ${{ vars.TENANT_DOMAIN }}

      - name: Purge CI-Produced Folders for Sites
        timeout-minutes: 30
        if: always()
        uses: ./.github/actions/purge-m365-data
        with:
          site: ${{ vars.CORSO_M365_TEST_SITE_URL }}
          folder-prefix: ${{ env.RESTORE_DEST_PFX }}
          libraries: ${{ vars.CORSO_M365_TEST_SITE_LIBRARIES }}
          older-than: ${{ env.NOW }}
          azure-client-id: ${{ secrets[needs.SetM365App.outputs.client_id_env] }}
          azure-client-secret: ${{ secrets[needs.SetM365App.outputs.client_secret_env] }}
          azure-tenant-id: ${{ secrets.TENANT_ID }}
          m365-admin-user: ${{ secrets.M365_TENANT_ADMIN_USER }}
          m365-admin-password: ${{ secrets.M365_TENANT_ADMIN_PASSWORD }}
          azure-pnp-client-id: ${{ secrets.AZURE_PNP_CLIENT_ID }}
          azure-pnp-client-cert: ${{ secrets.AZURE_PNP_CLIENT_CERT }}
          tenant-domain: ${{ vars.TENANT_DOMAIN }}

      ##########################################################################################################################################

      # Repository commands

      - name: Version Test
        timeout-minutes: 10
        run: |
          ./corso --version | grep -c 'Corso version:'

      - name: Repo init test
        timeout-minutes: 10
        id: repo-init
        run: |
          set -euo pipefail
          prefix=$(date +"%Y-%m-%d-%T")
          echo -e "\nRepo init test\n" >> ${{ env.CORSO_LOG_FILE }}
          ./corso repo init s3 \
            --no-stats \
            --hide-progress \
            --prefix $prefix \
            --bucket ${{ secrets.CI_TESTS_S3_BUCKET }} \
            2>&1 | tee ${{ env.CORSO_LOG_DIR }}/gotest-repo-init.log

          if ! grep -q 'Initialized a S3 repository within bucket' ${{ env.CORSO_LOG_DIR }}/gotest-repo-init.log
          then
            echo "Repo could not be initialized"
            exit 1
          fi

          echo result="$prefix" >> $GITHUB_OUTPUT

      - name: Repo connect test
        timeout-minutes: 10
        run: |
          set -euo pipefail
          echo -e "\nRepo connect test\n" >> ${{ env.CORSO_LOG_FILE }}
          ./corso repo connect s3 \
            --no-stats \
            --hide-progress \
            --prefix ${{ steps.repo-init.outputs.result }} \
            --bucket ${{ secrets.CI_TESTS_S3_BUCKET }} \
            2>&1 | tee ${{ env.CORSO_LOG_DIR }}/gotest-repo-connect.log

          if ! grep -q 'Connected to S3 bucket' ${{ env.CORSO_LOG_DIR }}/gotest-repo-connect.log
          then
            echo "Repo could not be connected"
            exit 1
          fi

      # Run maintenance on an empty repo just to make sure the command still
      # works.
      - name: Repo maintenance test
        timeout-minutes: 30
        run: |
          set -euo pipefail
          echo -e "\nRepo maintenance test\n" >> ${{ env.CORSO_LOG_FILE }}
          ./corso repo maintenance \
            --no-stats \
            --hide-progress \
            --mode complete \
            2>&1 | tee ${{ env.CORSO_LOG_DIR }}/gotest-repo-maintenance.log

      ##########################################################################################################################################

      # Exchange

      # generate new entries to roll into the next load test
      # only runs if the test was successful
      - name: Exchange - Create new data
        timeout-minutes: 30
        working-directory: ./src/cmd/factory
        run: |
          go run . exchange emails \
          --user ${{ env.TEST_USER }} \
          --tenant ${{ secrets.TENANT_ID }} \
          --destination ${{ env.RESTORE_DEST_PFX }}${{ steps.repo-init.outputs.result }} \
          --count 4

      - name: Exchange - Backup
        timeout-minutes: 30
        id: exchange-backup
        uses: ./.github/actions/backup-restore-test
        with:
          service: exchange
          kind: first-backup
          backup-args: '--mailbox "${{ env.TEST_USER }}" --data "email"'
          restore-args: "--email-folder ${{ env.RESTORE_DEST_PFX }}${{ steps.repo-init.outputs.result }}"
          restore-container: "${{ env.RESTORE_DEST_PFX }}${{ steps.repo-init.outputs.result }}"
          log-dir: ${{ env.CORSO_LOG_DIR }}
          with-export: true

      - name: Exchange - Incremental backup
        timeout-minutes: 30
        id: exchange-backup-incremental
        uses: ./.github/actions/backup-restore-test
        with:
          service: exchange
          kind: incremental
          backup-args: '--mailbox "${{ env.TEST_USER }}" --data "email"'
          restore-args: "--email-folder ${{ env.RESTORE_DEST_PFX }}${{ steps.repo-init.outputs.result }}"
          restore-container: "${{ env.RESTORE_DEST_PFX }}${{ steps.repo-init.outputs.result }}"
          backup-id: ${{ steps.exchange-backup.outputs.backup-id }}
          log-dir: ${{ env.CORSO_LOG_DIR }}
          with-export: true

      - name: Exchange - Non delta backup
        timeout-minutes: 30
        id: exchange-backup-non-delta
        uses: ./.github/actions/backup-restore-test
        with:
          service: exchange
          kind: non-delta
          backup-args: '--mailbox "${{ env.TEST_USER }}" --data "email" --disable-delta'
          restore-args: "--email-folder ${{ env.RESTORE_DEST_PFX }}${{ steps.repo-init.outputs.result }}"
          restore-container: "${{ env.RESTORE_DEST_PFX }}${{ steps.repo-init.outputs.result }}"
          backup-id: ${{ steps.exchange-backup.outputs.backup-id }}
          log-dir: ${{ env.CORSO_LOG_DIR }}
          with-export: true

      - name: Exchange - Incremental backup after non-delta
        timeout-minutes: 30
        id: exchange-backup-incremental-after-non-delta
        uses: ./.github/actions/backup-restore-test
        with:
          service: exchange
          kind: non-delta-incremental
          backup-args: '--mailbox "${{ env.TEST_USER }}" --data "email"'
          restore-args: "--email-folder ${{ env.RESTORE_DEST_PFX }}${{ steps.repo-init.outputs.result }}"
          restore-container: "${{ env.RESTORE_DEST_PFX }}${{ steps.repo-init.outputs.result }}"
          backup-id: ${{ steps.exchange-backup.outputs.backup-id }}
          log-dir: ${{ env.CORSO_LOG_DIR }}
          with-export: true

      ##########################################################################################################################################

      # Onedrive

      # generate new entries for test
      - name: OneDrive - Create new data
        id: new-data-creation-onedrive
        timeout-minutes: 30
        working-directory: ./src/cmd/factory
        run: |
          suffix=$(date +"%Y-%m-%d_%H-%M-%S")

          go run . onedrive files  \
            --user ${{ env.TEST_USER }} \
            --secondaryuser  ${{ env.CORSO_SECONDARY_M365_TEST_USER_ID }} \
            --tenant ${{ secrets.TENANT_ID }} \
            --destination ${{ env.RESTORE_DEST_PFX }}$suffix \
            --count 4

          echo result="${suffix}" >> $GITHUB_OUTPUT

      - name: OneDrive - Backup
        id: onedrive-backup
        timeout-minutes: 30
        uses: ./.github/actions/backup-restore-test
        with:
          service: onedrive
          kind: first-backup
          backup-args: '--user "${{ env.TEST_USER }}"'
          restore-args: "--folder ${{ env.RESTORE_DEST_PFX }}${{ steps.new-data-creation-onedrive.outputs.result }}"
          restore-container: "${{ env.RESTORE_DEST_PFX }}${{ steps.new-data-creation-onedrive.outputs.result }}"
          log-dir: ${{ env.CORSO_LOG_DIR }}
          with-export: true

      # generate some more enteries for incremental check
      - name: OneDrive - Create new data (for incremental)
        timeout-minutes: 30
        working-directory: ./src/cmd/factory
        run: |
          go run . onedrive files  \
            --user ${{ env.TEST_USER }} \
            --secondaryuser  ${{ env.CORSO_SECONDARY_M365_TEST_USER_ID }} \
            --tenant ${{ secrets.TENANT_ID }} \
            --destination ${{ env.RESTORE_DEST_PFX }}${{ steps.new-data-creation-onedrive.outputs.result }} \
            --count 4

      - name: OneDrive - Incremental backup
        id: onedrive-incremental
        timeout-minutes: 30
        uses: ./.github/actions/backup-restore-test
        with:
          service: onedrive
          kind: incremental
          backup-args: '--user "${{ env.TEST_USER }}"'
          restore-args: "--folder ${{ env.RESTORE_DEST_PFX }}${{ steps.new-data-creation-onedrive.outputs.result }}"
          restore-container: "${{ env.RESTORE_DEST_PFX }}${{ steps.new-data-creation-onedrive.outputs.result }}"
          log-dir: ${{ env.CORSO_LOG_DIR }}
          with-export: true

      ##########################################################################################################################################

      # Sharepoint Library

      # generate new entries for test
      - name: SharePoint - Create new data
        id: new-data-creation-sharepoint
        timeout-minutes: 30
        working-directory: ./src/cmd/factory
        run: |
          suffix=$(date +"%Y-%m-%d_%H-%M-%S")

          go run . sharepoint files  \
            --site ${{ vars.CORSO_M365_TEST_SITE_URL }} \
            --user ${{ env.TEST_USER }} \
            --secondaryuser  ${{ env.CORSO_SECONDARY_M365_TEST_USER_ID }} \
            --tenant ${{ secrets.TENANT_ID }} \
            --destination ${{ env.RESTORE_DEST_PFX }}$suffix \
            --count 4

          echo result="${suffix}" >> $GITHUB_OUTPUT

      - name: SharePoint - Backup
        id: sharepoint-backup
        timeout-minutes: 30
        uses: ./.github/actions/backup-restore-test
        with:
          service: sharepoint
          kind: first-backup
          backup-args: '--site "${{ vars.CORSO_M365_TEST_SITE_URL }}" --data libraries'
          restore-args: "--folder ${{ env.RESTORE_DEST_PFX }}${{ steps.new-data-creation-sharepoint.outputs.result }}"
          restore-container: "${{ env.RESTORE_DEST_PFX }}${{ steps.new-data-creation-sharepoint.outputs.result }}"
          log-dir: ${{ env.CORSO_LOG_DIR }}
          with-export: true
          category: libraries

      # generate some more enteries for incremental check
      - name: SharePoint - Create new data (for incremental)
        timeout-minutes: 30
        working-directory: ./src/cmd/factory
        run: |
          go run . sharepoint files  \
            --site ${{ vars.CORSO_M365_TEST_SITE_URL }} \
            --user ${{ env.TEST_USER }} \
            --secondaryuser  ${{ env.CORSO_SECONDARY_M365_TEST_USER_ID }} \
            --tenant ${{ secrets.TENANT_ID }} \
            --destination ${{ env.RESTORE_DEST_PFX }}${{ steps.new-data-creation-sharepoint.outputs.result }} \
            --count 4

      - name: SharePoint - Incremental backup
        id: sharepoint-incremental
        timeout-minutes: 30
        uses: ./.github/actions/backup-restore-test
        with:
          service: sharepoint
          kind: incremental
          backup-args: '--site "${{ vars.CORSO_M365_TEST_SITE_URL }}" --data libraries'
          restore-args: "--folder ${{ env.RESTORE_DEST_PFX }}${{ steps.new-data-creation-sharepoint.outputs.result }}"
          restore-container: "${{ env.RESTORE_DEST_PFX }}${{ steps.new-data-creation-sharepoint.outputs.result }}"
          log-dir: ${{ env.CORSO_LOG_DIR }}
          with-export: true
          category: libraries

      ##########################################################################################################################################

      # Sharepoint Lists

      # generate new entries for test
      # The `awk | tr | sed` command chain is used to get a comma separated list of SharePoint list names.
      - name: SharePoint Lists - Create new data
        id: new-data-creation-sharepoint-lists
        timeout-minutes: 30
        working-directory: ./src/cmd/factory
        run: |
          suffix=$(date +"%Y-%m-%d_%H-%M-%S")

          go run . sharepoint lists  \
            --site ${{ vars.CORSO_M365_TEST_SITE_URL }} \
            --user ${{ env.TEST_USER }} \
            --secondaryuser  ${{ env.CORSO_SECONDARY_M365_TEST_USER_ID }} \
            --tenant ${{ secrets.TENANT_ID }} \
            --destination ${{ env.RESTORE_DEST_PFX }}$suffix \
            --count 4 |
            awk 'NR > 1 {print $2}' | tr '\n' ',' | sed -e 's/,$//' -e 's/^/result=/' |
            tee $GITHUB_OUTPUT
      # Extracts the common prefix for the Sharepoint list names.
      - name: SharePoint Lists - Store restore container
        id: sharepoint-lists-store-restore-container
        run: |
          echo ${{ steps.new-data-creation-sharepoint-lists.outputs.result }} |
          cut -d',' -f1 |
          cut -d'_' -f1,2,3,4,5 |
          sed -e 's/^/result=/' |
          tee $GITHUB_OUTPUT

      - name: SharePoint Lists - Backup
        id: sharepoint-lists-backup
        timeout-minutes: 30
        uses: ./.github/actions/backup-restore-test
        with:
          service: sharepoint
          kind: first-backup-lists
          backup-args: '--site "${{ vars.CORSO_M365_TEST_SITE_URL }}" --data lists'
          restore-args: "--list ${{ steps.new-data-creation-sharepoint-lists.outputs.result }} --destination Corso_Test_Sanity_Restore_$(date +'%Y%m%d_%H%M%S')"
          export-args: "--list ${{ steps.new-data-creation-sharepoint-lists.outputs.result }}"
          restore-container: "${{ steps.sharepoint-lists-store-restore-container.outputs.result }}"
          log-dir: ${{ env.CORSO_LOG_DIR }}
          with-export: true
          category: lists
          on-collision: copy

      # generate some more enteries for incremental check
      - name: SharePoint Lists - Create new data (for incremental)
        id: inc-data-creation-sharepoint-lists
        timeout-minutes: 30
        working-directory: ./src/cmd/factory
        run: |
          suffix=$(date +"%Y-%m-%d_%H-%M-%S")

          go run . sharepoint lists  \
            --site ${{ vars.CORSO_M365_TEST_SITE_URL }} \
            --user ${{ env.TEST_USER }} \
            --secondaryuser  ${{ env.CORSO_SECONDARY_M365_TEST_USER_ID }} \
            --tenant ${{ secrets.TENANT_ID }} \
            --destination ${{ env.RESTORE_DEST_PFX }}$suffix \
            --count 4 |
            awk 'NR > 1 {print $2}' | tr '\n' ',' | sed -e 's/,$//' -e 's/^/result=/' |
            tee $GITHUB_OUTPUT

      - name: SharePoint Lists - Store restore container (for incremental)
        id: sharepoint-lists-store-restore-container-inc
        run: |
          echo ${{ steps.inc-data-creation-sharepoint-lists.outputs.result }} |
          cut -d',' -f1 |
          cut -d'_' -f1,2,3,4,5 |
          sed -e 's/^/result=/' |
          tee $GITHUB_OUTPUT

      - name: SharePoint Lists - Incremental backup
        id: sharepoint-lists-incremental
        timeout-minutes: 30
        uses: ./.github/actions/backup-restore-test
        with:
          service: sharepoint
          kind: incremental-lists
          backup-args: '--site "${{ vars.CORSO_M365_TEST_SITE_URL }}" --data lists'
          restore-args: "--list ${{ steps.inc-data-creation-sharepoint-lists.outputs.result }},${{ steps.new-data-creation-sharepoint-lists.outputs.result }} --destination Corso_Test_Sanity_Restore_$(date +'%Y%m%d_%H%M%S')"
          export-args: "--list ${{ steps.inc-data-creation-sharepoint-lists.outputs.result }},${{ steps.new-data-creation-sharepoint-lists.outputs.result }}"
          restore-container: "${{ steps.sharepoint-lists-store-restore-container-inc.outputs.result }},${{ steps.sharepoint-lists-store-restore-container.outputs.result }}"
          log-dir: ${{ env.CORSO_LOG_DIR }}
          with-export: true
          category: lists
          on-collision: copy

      ##########################################################################################################################################

      # Groups and Teams

      # generate new entries for test
      - name: Groups - Create new data
        id: new-data-creation-groups
        timeout-minutes: 30
        working-directory: ./src/cmd/factory
        run: |
          suffix=$(date +"%Y-%m-%d_%H-%M-%S")

          go run . sharepoint files  \
            --site ${{ vars.CORSO_M365_TEST_GROUPS_SITE_URL }} \
            --user ${{ env.TEST_USER }} \
            --secondaryuser  ${{ env.CORSO_SECONDARY_M365_TEST_USER_ID }} \
            --tenant ${{ secrets.TENANT_ID }} \
            --destination ${{ env.RESTORE_DEST_PFX }}$suffix \
            --count 4

          echo result="${suffix}" >> $GITHUB_OUTPUT

      - name: Groups - Backup
        id: groups-backup
        timeout-minutes: 30
        uses: ./.github/actions/backup-restore-test
        with:
          service: groups
          kind: first-backup
          backup-args: '--group "${{ vars.CORSO_M365_TEST_TEAM_ID }}" --data messages,libraries'
          restore-container: "${{ env.RESTORE_DEST_PFX }}${{ steps.new-data-creation-groups.outputs.result }}"
          log-dir: ${{ env.CORSO_LOG_DIR }}
          with-export: true

      # generate some more entries for incremental check
      - name: Groups - Create new data (for incremental)
        timeout-minutes: 30
        working-directory: ./src/cmd/factory
        run: |
          go run . sharepoint files  \
            --site ${{ vars.CORSO_M365_TEST_GROUPS_SITE_URL }} \
            --user ${{ env.TEST_USER }} \
            --secondaryuser  ${{ env.CORSO_SECONDARY_M365_TEST_USER_ID }} \
            --tenant ${{ secrets.TENANT_ID }} \
            --destination ${{ env.RESTORE_DEST_PFX }}${{ steps.new-data-creation-groups.outputs.result }} \
            --count 4

      - name: Groups - Incremental backup
        id: groups-incremental
        timeout-minutes: 30
        uses: ./.github/actions/backup-restore-test
        with:
          service: groups
          kind: incremental
          backup-args: '--group "${{ vars.CORSO_M365_TEST_TEAM_ID }}" --data messages,libraries'
          restore-args: '--site "${{ vars.CORSO_M365_TEST_GROUPS_SITE_URL }}" --folder ${{ env.RESTORE_DEST_PFX }}${{ steps.new-data-creation-groups.outputs.result }}'
          restore-container: "${{ env.RESTORE_DEST_PFX }}${{ steps.new-data-creation-groups.outputs.result }}"
          log-dir: ${{ env.CORSO_LOG_DIR }}
          with-export: true

      ##########################################################################################################################################

      # Logging & Notifications

      # Upload the original go test output as an artifact for later review.
      - name: Upload test log
        if: always()
        uses: actions/upload-artifact@v4
        with:
          name: sanity-test-log
          path: ${{ env.CORSO_LOG_DIR }}/*
          if-no-files-found: error
          retention-days: 14

      - name: Notify failure in teams
        if: failure()
        uses: ./.github/actions/teams-message
        with:
          msg: "[CORSO FAILED] Sanity Tests"
          teams_url: ${{ secrets.TEAMS_CORSO_CI_WEBHOOK_URL }}