Redis vss
๐ REDIS is used to support fast vector search. This โก๏ธaccelerates machine learning workflows that require vector similarity search using foundational or fine-tuned models. Each project or project system has its own Redis instance, which supports isolation of the data.

To add a new Redis instance:
- Add support for a new volume, Redis service, and port to the compose.yml file in the deploy-ai repository: playbooks/files/fastapi/vss.
For example, with a new name and port, e.g. redis-sinker on port 6383:
services:
fastapi:
image: mbari/fastapi-vss:${GIT_VERSION}
container_name: fastapi-vss
healthcheck:
test: [ "CMD", "curl", "-f", "http://127.0.0.1:8000/health" ]
env_file:
- .env.prod
ports:
- "8000:80"
volumes:
- /opt/deployment:/opt/deployment
user: ${MLDEVOPS_GID}:${MLDEVOPS_UID}
restart: always
depends_on:
- redis-uav
- redis-cfe
- redis-bio
- redis-i2map
- redis-sinker , ๐ add this line
redis-planktivore: , ๐ add this block
env_file:
- .env.prod
command:
- /bin/sh
- -c
- redis-stack-server --port 6383 --appendonly yes --appendfsync everysec --requirepass "$${REDIS_PASSWD:?REDIS_PASSWD variable is not set}"
image: redis/redis-stack-server
container_name: redis-sinker
ports:
- "6383:6383"
volumes:
- redis-sinker-data:/var/lib/redis-stack
restart: always
networks:
- vss
volumes:
redis-uav-data:
redis-cfe-data:
redis-bio-data:
redis-i2map-data:
redis-sinker-data: ๐ add this line
networks:
vss:
-
Create a new config file in the
configdirectory in the aidata repository, e.g.config_planktivore.yml -
Initialize the database with the aidata command
db resestwith that config file, e.g.aidata db reset --config aidata/config/config_planktivore_lm.yml --redis-password <password>
Tip
Logs can be seen on the devops mount at smb://thalassa.shore.mbari.org/DevOps This can be useful for debugging issues with services in docker images.
๐๏ธ Updated: 2025-07-07