Migrate Feature Vectors to Different Neural Network Model

Tip

Do not hesitate to contact our experts on migration by support@ntechlab.com.

Important

In the case, you are doing migration as part of the system update to a newer version, complete the update first. Only after that can you proceed to the migration.

This section is about how to migrate object feature vectors to another neural network model.

Do the following:

  1. Create a backup of the Tarantool-based feature vector database in any directory of your choice, for example, /etc/findface_dump.

    sudo mkdir -p /etc/findface_dump
    sudo cd /etc/findface_dump
    sudo findface-storage-api-dump -config /etc/findface-sf-api.ini
    
  2. Stop the findface-sf-api service.

    sudo systemctl stop findface-sf-api.service
    
  3. Create new shards that will host regenerated feature vectors.

    1. Open the /etc/tarantool/instances.available/ directory and find out the number of shards by counting the number of configuration files shard-*.lua.

      Note

      There are four shards in the example below.

      cd /etc/tarantool/instances.available/
      
      ls -l
      
      shard-001.lua
      shard-002.lua
      shard-003.lua
      shard-004.lua
      
    2. Create the same number of new shards by copying the configuration files shard-*.lua.

      Note

      For convenience, the second digit in the new names is 1: shard-01*.lua.

      sudo cp shard-001.lua shard-011.lua
      sudo cp shard-002.lua shard-012.lua
      sudo cp shard-003.lua shard-013.lua
      sudo cp shard-004.lua shard-014.lua
      
    3. Modify the following lines in each new shard’s configuration file, subject to its name (shard-011, shard-012, etc., in our example):

      Old value

      New value

      listen = ‘127.0.0.1:32001’

      Listen = ‘127.0.0.1:32011’

      vinyl_dir = ‘/opt/ntech/var/lib/tarantool/shard-001’

      vinyl_dir = ‘/opt/ntech/var/lib/tarantool/shard-011’

      work_dir = ‘/opt/ntech/var/lib/tarantool/shard-001’

      work_dir = ‘/opt/ntech/var/lib/tarantool/shard-011’

      memtx_dir = ‘/opt/ntech/var/lib/tarantool/shard-001/snapshots’

      memtx_dir = ‘/opt/ntech/var/lib/tarantool/shard-011/snapshots’

      wal_dir = ‘/opt/ntech/var/lib/tarantool/shard-001/xlogs’

      wal_dir = ‘/opt/ntech/var/lib/tarantool/shard-011/xlogs’

      FindFace.start(“127.0.0.1”, 8101, {

      FindFace.start(“127.0.0.1”, 8111, {

    4. Create symbolic links to the new shards.

      cd /etc/tarantool/instances.enabled/
      
      sudo ln -s /etc/tarantool/instances.available/shard-01*.lua /etc/tarantool/instances.enabled/
      
    5. Create directories that will host files of the new shards. Assign permissions for the created directories.

      cd /opt/ntech/var/lib/tarantool/
      
      mkdir -p shard-01{1..4}/{index,snapshots,xlogs}
      
      chown tarantool:tarantool shard-01* shard-01*/*
      
  4. Open the /etc/findface-extraction-api configuration file and replace the extraction models with the new ones in the body_emben, car_emben, and face_emben parameters, depending on the object types to migrate.

    sudo vi /etc/findface-extraction-api.ini
    
    extractors:
      ...
      models:
        ...
        body_emben: pedrec/<new_model_body>.gpu.fnk
        ...
        car_emben: carrec/<new_model_car>.gpu.fnk
        ...
        face_emben: face/<new_model_face>.gpu.fnk
    

    Restart the findface-extraction-api service.

    sudo systemctl restart findface-extraction-api.service
    
  5. Start the new shards.

    for i in {11..14}; do sudo systemctl start [email protected]$i; done
    
  6. Create a configuration file with migration settings <migration.ini> based on the example below.

    extraction-api:
      timeouts:
        connect: 5s
        response_header: 30s
        overall: 35s
        idle_connection: 0s
      extraction-api: http://127.0.0.1:18666
    storage-api-from: # current location of the gallery
      timeouts:
        connect: 5s
        response_header: 30s
        overall: 35s
        idle_connection: 10s
      max-idle-conns-per-host: 20
      shards:
        - master: http://127.0.0.1:8101/v2/
          slave: ""
        - master: http://127.0.0.1:8102/v2/
          slave: ""
        - master: http://127.0.0.1:8103/v2/
          slave: ""
        - master: http://127.0.0.1:8104/v2/
          slave: ""
    storage-api-to:
      timeouts:
        connect: 5s
        response_header: 30s
        overall: 35s
        idle_connection: 10s
      max-idle-conns-per-host: 20
      shards:
        - master: http://127.0.0.1:8111/v2/
          slave: ""
        - master: http://127.0.0.1:8112/v2/
          slave: ""
        - master: http://127.0.0.1:8113/v2/
          slave: ""
        - master: http://127.0.0.1:8114/v2/
          slave: ""
    workers_num: 3
    faces_limit: 100
    extraction_batch_size: 8
    normalized_storage:
      type: webdav
      enabled: True
      webdav:
        upload-url: http://127.0.0.1:3333/uploads/
      s3:
        endpoint: ""
        bucket-name: ""
        access-key: ""
        secret-access-key: ""
        secure: False
        region: ""
        public-url: ""
        operation-timeout: 30
    

    In the storage-api-from section, specify the old shards to migrate the data from.

    storage-api-from: # current location of the gallery
      ...
      shards:
        - master: http://127.0.0.1:8101/v2/
          slave: ""
        - master: http://127.0.0.1:8102/v2/
          slave: ""
        - master: http://127.0.0.1:8103/v2/
          slave: ""
         - master: http://127.0.0.1:8104/v2/
      ...
    

    In the storage-api-to section, specify the new shards that will host migrated data.

    storage-api-to:
      ...
      shards:
        - master: http://127.0.0.1:8111/v2/
          slave: ""
        - master: http://127.0.0.1:8112/v2/
          slave: ""
        - master: http://127.0.0.1:8113/v2/
          slave: ""
        - master: http://127.0.0.1:8114/v2/
          slave: ""
      ...
    
  7. Launch the findface-sf-api-migrate utility with the -config option and provide the <migration.ini> configuration file.

    findface-sf-api-migrate -config migration.ini
    

    Note

    The migration process can take up a significant amount of time if there are many events and cards in the system.

  8. After the migration is complete, stop the old shards and disable their autostart in OS (do not remove them).

    for i in {01..04}; do sudo systemctl stop [email protected]$i.service ; done
    
    for i in {01..04}; do sudo systemctl disable [email protected]$i.service ; done
    
  9. Open the /etc/findface-sf-api.ini configuration file and adjust the shards ports, subject to the new shards settings. Restart the findface-sf-api service.

    sudo vi /etc/findface-sf-api.ini
    
    shards:
      - master: http://127.0.0.1:8111/v2/
        slave: ""
      - master: http://127.0.0.1:8112/v2/
        slave: ""
      - master: http://127.0.0.1:8113/v2/
        slave: ""
      - master: http://127.0.0.1:8114/v2/
        slave: ""
    
    sudo systemctl start findface-sf-api.service
    
  10. Import the new database structure from the tnt_schema.lua file.

    sudo findface-security make_tnt_schema | sudo tee /etc/findface-security/tnt_schema.lua
    
  11. Migrate the main database architecture from FindFace Multi to PostgreSQL, re-create predefined user roles and the first administrator.

    sudo findface-security migrate
    sudo findface-security create_groups
    sudo findface-security create_default_user
    
  12. Restart the services.

    sudo systemctl restart findface-security.service
    sudo systemctl restart findface-extraction-api findface-video-worker* findface-video-manager findface-sf-api
    
  13. Migrate clusters as well if this functionality is enabled in your system. To do so, execute the following command:

    Note

    List the object types to migrate as the command options: --face, --body, --car.

    sudo findface-security migrate_clusters --face --body --car --use-best-event --use-thumbnail --force-clusterin
    

    As a result, the system will regenerate feature vectors for the existing cluster events and automatically launch the scheduled clustering to rebuild clusters.