Bisque Docker Environment Setup Instructions¶

Docker/Project Source¶

• Docker Hub Bisque Dev Image
• UCSB-VRL/bisque-stable Github
• Bique Bioimage Google Groups

Installation¶

Pre-requisite¶

• Install docker (light weight VM) on you laptop: https://docs.docker.com/install/linux/docker-ce/ubuntu/
• Instructions on installing bisque using docker https://github.com/UCSB-VRL/bisque/README.md

Create the Docker Image¶

Look into the UCSB-VRL/bisque-build repository on how to create Docker container images for development

• The container image has bisque server, docker & virtual environment pre-installed
• The child containers that will run modules, will use the host network since we are working with local IP addresses which are not resolved on development network environments.

Run Environment¶

• Setup folders & pull code (Only for modules since Bisque server codebase is already part of the container image being used in next step)

mkdir ws && cd ws && git clone https://github.com/UCSB-VRL/bisque
mkdir container-modules container-data container-config && cp -r bisque/modules/* container-modules/

• Start Docker and mount directories for run

# Docker Run
docker run --name bisque-dev --rm -p 8080:8080 -p 27000:27000 \
 -v $(pwd)/container-modules:/source/modules \
 -v $(pwd)/container-data:/source/data \
 -v $(pwd)/container-config:/source/config \
 -v /var/run/docker.sock:/var/run/docker.sock \
 --ipc=host \
 --net=host \
 'vishwakarmarhl/ucsb-bisque05-svc:dev'

Alternately if you want an interactive log for docker run use the below command

docker run -it -p 8080:8080 -p 27000:27000 \
 -v $(pwd):/ws \
 -v /var/run/docker.sock:/var/run/docker.sock \
 --mount source=bqvol,target=/ws/bqvol \
 --ipc=host \
 --net=host \
 ucsb-bisque05-svc:dev

• Check Container state

bisque@ubuntu:~$ docker ps
CONTAINER ID        IMAGE                                  COMMAND                  CREATED             STATUS              PORTS                                              NAMES
a1162677d66a        vishwakarmarhl/ucsb-bisque05-svc:dev   "/builder/run-bisque…"   13 minutes ago      Up 13 minutes       0.0.0.0:8080->8080/tcp, 0.0.0.0:27000->27000/tcp   bisque-dev

• Stop Docker

docker stop $(docker ps -a -q --filter ancestor=vishwakarmarhl/ucsb-bisque05-svc:dev --format="{{.ID}}")

• This will start the bisque docker/server
  • Client Service http://0.0.0.0:8080
  • Engine Service http://0.0.0.0:8080/engine_service
  • Module Service http://0.0.0.0:8080/module_service
  • Features http://0.0.0.0:8080/features

• Since we want to develop we should use the local IP address for all accesses

  • Client Service http://192.168.0.14:8080/
  • Engine Service http://192.168.0.14:8080/engine_service

• Use "admin:admin" to authenticate into this local environment

• To manipulate and check logs, connect to docker bash

bisque-host$ docker exec -t -i bisque-dev /bin/bash

• Advanced environment for working with database URI as
  • sqlite:///data/bisque.db
  • postgresql://rahul:rahul@localhost/bqmurks58 or postgresql://dbhost:5432/bisque

docker run --name bisque-dev --rm -p 8080:8080 -p 27000:27000 \
-e BISQUE_DBURL=postgresql://rahul:rahul@localhost/bqmurks58 \
'vishwakarmarhl/ucsb-bisque05-svc:dev'

Develop in Docker (Work In Progress !!)¶

Run the containerized bisque server as described in the first half of this document

• Use the dev docker image docker vishwakarmarhl/ucsb-bisque05-svc
• Bash into the docker container and develop

docker exec -it bisque-dev /bin/bash

• Activate environment and Test/Build a Module

root@9362d1f8bf12:/source# cd modules/MaskRCNN
root@9362d1f8bf12:/source/modules/MaskRCNN# source /usr/lib/bisque/bin/activate
root@9362d1f8bf12:/source/modules/MaskRCNN# python setup.py

• Verify Bisque Services
• Now you can add the MaskRCNN module from the manager interface and run a test
• In this test a MEX-ID will be assigned to this module run and corresponding code will be dumped in the staging folder
• Look for "docker_run" script in the "staging/" folder. It has the following commands which you can use to run and debug your codebase
• Script first creates the parameterized container command

docker pull biodev.ece.ucsb.edu:5000/tflow-mrcnn-mod-v1
CONTAINER=$(docker create --ipc=host biodev.ece.ucsb.edu:5000/tflow-mrcnn-mod-v1  python PythonScriptWrapper.py http://192.168.0.14:8080/data_service/00-s3KxZhuvBt8vhER5RD7wdX 15 0.05 http://192.168.0.14:8080/module_service/mex/00-wRq68hGv5HTtr6mBsfM4CG admin:00-wRq68hGv5HTtr6mBsfM4CG)

• Thereafter it executes the parameterized container command

docker start $CONTAINER
MODULE_RETURN=$(docker wait  $CONTAINER)
docker logs $CONTAINER

• In your test, while still in the staging folder you can directly use the "docker_run" script to run your module as follows

sh docker_run python PythonScriptWrapper.py \
  http://192.168.0.14:8080/data_service/00-s3KxZhuvBt8vhER5RD7wdX \
  15 0.05 \
  http://192.168.0.14:8080/module_service/mex/00-wRq68hGv5HTtr6mBsfM4CG \
  admin:00-wRq68hGv5HTtr6mBsfM4CG)

Notes: Look into the README files on the modules for more commands on test/debug

• MaskRCNN - README from Github
• Deep Planteome - Module Instructions