EVSRESTAPI

Information on the build and deployment process for the EVSRESTAPI project

Prerequisites

• Install Docker and ensure it is configured to allow (Docker -> Settings -> Resources)
  • Memory = 6G
  • Swap = 1G
• Clone the project - https://github.com/NCIEVS/evsrestapi
• Choose a local directory $dir (e.g. c:/evsrestapi)
• mkdir -p $dir/elasticsearch/data
• Download latest ThesaurusInf_*OWL.zip, unpack to $dir/ThesaurusInferred.owl (see https://evs.nci.nih.gov/ftp1/NCI_Thesaurus/)

Steps for Loading Data and Indexes Locally

• Launch Stardog and load NCI Thesaurus data - (see Stardog Resources)

• Launch Elasticsearch docker container In a terminal/Cygwin window, run the following to have an elasticsearch instance running. Keep this window open to keep the server running.

  docker pull docker.elastic.co/elasticsearch/elasticsearch:6.7.0
  # Choose a directory for your elasticsearch data to live
  dir=c:/evsrestapi/elasticsearch/data
  docker run -p 9200:9200 -v "$dir":/usr/share/elasticsearch/data  -e "discovery.type=single-node" -e ES_JAVA_OPTS="-Xms3g -Xmx4g"  docker.elastic.co/elasticsearch/elasticsearch:6.7.0
  

• Load/Compute Indexes - Run from the “elasticsearch/scripts” folder of the cloned https://github.com/NCIEVS/evsrestapi repo.

  # From the root of cloned https://github.com/NCIEVS/evsrestapi
  cd elasticsearch/scripts
  
  # Configure .env file
  # NOTE: host.docker.internal refers to "localhost" of the environment running docker, 
  #       an easy way to avoid having to network containers together.
  cat > .env << EOF
  API_ENDPOINT=http://host.docker.internal:8080/api/v1/concept/
  CONCEPT_OUTPUT_DIR=/tmp/
  CONCEPT_INPUT_DIR=/tmp/
  ES_HOST=host.docker.internal
  ES_PORT=9200
  INDEX_MAPPING_FILE=../Mapping/concept_flat_full_nested_bylabel.json
  LOG_DIRECTORY=/tmp/
  MULTI_PROCESSING_POOL=2
  NAMED_GRAPH=http://NCI_T
  SPARQL_ENDPOINT=http://host.docker.internal:5820/NCIT2/query
  STARDOG_USERNAME=admin
  STARDOG_PASSWORD=admin
  EOF
  

• Run the python load script in the docker container (from the “elasticsearch” directory of the cloned project)

  # From the root of cloned https://github.com/NCIEVS/evsrestapi
  cd elasticsearch
  
  docker build -t evsrestapi:elasticsearch .
  docker run -it evsrestapi:elasticsearch
  (base) root@b63d5d1f4038:/elasticsearch# cd scripts/
  (base) root@b63d5d1f4038:/elasticsearch# ./bulkLoadES.py --index_name concept --download_only
  (base) root@b63d5d1f4038:/elasticsearch# ./bulkLoadES.py --index_name concept --no_download --delete_document
  (base) root@b63d5d1f4038:/elasticsearch# exit
  

Steps for Building and Running EVSRESTAPI locally

• Launch Stardog and Elasticsearch (as described above)

  • If loaded properly, the loaded artifacts should be persistent and you can take down and restart the docker processes and the data will still be there.
  • NOTE: both services must be loaded and running for the application tests to run properly

• Configure application

  • see src/main/resources/application-local.yml file for local setup (these settings should be suitable for local deployment)

• Build the application (MUST DO BEFORE RUNNING if using “external tools configuration”)

  • SPRING_PROFILES_ACTIVE=local ./gradlew clean build
  • Executable war file present in build/libs

• Run application in Eclipse (SpringBoot)

  • Click "Run" → "External Tools" → "External Tools Configurations"

  • Create a new entry under “Program” and configure it as follows:

    • location = <path to java executable, e.g. C:/Program Files/Java/jdk1.8.0_191/bin/java.exe>
    • working dir = <path to project, e.g. C:/Users/bcarl/Desktop/workspace/evsrestapi>
    • Arguments = command line args
      • -Xmx4096M - ensure enough memory usage
      • -Dspring.profiles.active=local - make sure to use application-local.yml
      • -jar *.war - point to the war file

  • Test that it’s up by looking for swagger docs: http://localhost:8080/swagger-ui.html#/