0

We need preloaded Weaviate Docker image for local development (the database is not big and can go in GIT). We need something like in Postgres ( preloading with test_dump)

COPY test_dump.sql /docker-entrypoint-initdb.d/

Current workaround is with the backup module

ENABLE_MODULES: 'text2vec-openai,backup-filesystem'

I have crated backup and mounted it with volume in docker-compose.But extra step is needed with this solution. When container is already running, the restore database needs to be done manually (via rest endpoint http://localhost:8090/v1/backups/filesystem/my-very-first-backup/restore).

Current docker-compose.yaml with backup-filesystem module

version: "3.7"
services:
  weaviate:
    image: semitechnologies/weaviate:1.17.3
    ports:
      - 8081:8080
    environment:
      QUERY_DEFAULTS_LIMIT: 25
      AUTHENTICATION_ANONYMOUS_ACCESS_ENABLED: 'true'
      PERSISTENCE_DATA_PATH: './data'
      ENABLE_MODULES: 'text2vec-openai,backup-filesystem'
      DEFAULT_VECTORIZER_MODULE: 'text2vec-openai'
      OPENAI_APIKEY: $OPEN_AI_TOKEN
      BACKUP_FILESYSTEM_PATH: '/var/lib/weaviate-backup'
      AUTOSCHEMA_ENABLED: 'false'
      CLUSTER_HOSTNAME: 'node1'
      LOG_LEVEL: 'debug'
    volumes:
      - ./weaviate_data:/var/lib/weaviate-backup

Looking for a clean solution where a new developer will run

docker-compose up

and preloaded Weaviate database will be up&running

Levijatanu
  • 371
  • 2
  • 17
  • 1
    It's actually not uncommon to store the index files from the `./weaviate_data:/var/lib/weaviate-backup` folder. If you mount this folder to another Weaviate, it will directly load the data into the DB. – Bob van Luijt Apr 14 '23 at 16:36

0 Answers0