安装

需求：

• R2base (convmind/r2base:only-es)
• Elasticsearch (convmind/model-as-service:es-7.5.0)
• soco-search-server (convmind/soco-search:0.1-server)
• soco-search-worker (convmind/soco-search:0.1-worker)
• soco-search-controller (convmind/soco-search:0.1-controller)
• soco-encoders (convmind/soco-encoders:0.1-visual-sparta / convmind/soco-encoders:0.1-grpc)
• MongoDB (mongo)
• RedisDB (redis)

通过docker-compose安装：

1. 参考 https://github.com/soco-ai/R2Base ， 设置R2base和 Elasticsearch

git clone https://github.com/soco-ai/R2Base
cd R2Base
sudo docker-compose -d docker-compose.yml up

2. 从github上克隆soco-search的项目

git clone https://github.com/soco-ai/soco-search.git 

3. docker/var.env中更改环境变量

# General Setting
REGION=US # {"US","CN"}
MODE=docker-compose # {"docker-compose", "k8s"}

# MONGODB
MONGODB_URL=mongodb://192.1.2.219:6005/soco  # Database for saving data 
META_MONGODB_URL=mongodb://192.1.2.219:6005/search # Database for meta data, mainly for dashboard
MONGODB_SSL=false  {true, false}

# REDISDB
REDIS_URL=redis://192.1.2.219:6004/0 
REDIS_BATCH_SIZE=100 
REDIS_Q_NAME=SOCO_SEARCH_API # the queue name
Q_SIZE=20 # The number of jobs in the redis queue to prevent the size limit of Redisdb. 

# Elasticsearch
ES_URL=https://elastic:socoes2020@192.1.2.219:30005 

# Encoders (soco-encoders)
GRPC_ENCODER_SERVER=192.1.2.219:6003 # GRPC Encoder Server 

# Workers (soco-search-worker)
N_WORKERS=4
OBJC_DISABLE_INITIALIZE_FORK_SAFETY=YES # To allow multi-processing

# Controller (soco-search-controller)
SLEEP_INTERVAL=5 # When there is no job or error, sleep for 5 seconds, in order to control the speed of indexing. 
DB_BATCH_SIZE=200 # the batch size 

# Dashboard (soco-search-dashboard)
NODE_MODULES_CACHE=false {true, false}  # use cache to speed up the build times.
BACKEND_URL=http://192.1.2.219:6001 # the same as the front-end url

# Server (soco-search-server)
SOCO_SEARCH_SERVER=http://192.1.2.219:6002
R2BASE_URL=http://192.1.2.219:8000 

4. 运行 docker-compose

sudo docker-compose -f docker/docker-compose-soco-search.yml up