Skip to content

Latest commit

 

History

History
93 lines (56 loc) · 2.23 KB

running_the_client_and_server.md

File metadata and controls

93 lines (56 loc) · 2.23 KB
Raw

Setting up the Client and Server

In Short

from the modeldb/ directory

Server

cd server
cd codegen
./gen_sqlite.sh
cd ..
./start_server.sh

Client

cd client/scala/libs/spark.ml ./build_client.sh

The JAR file is then in:
target/scala-2.11/ml.jar

The Server, Step by Step

The server is in the modeldb/server/ directory.

From here on out, all commands will assume that you are in the /server directory.

cd server

Environment Variables

Make sure the following variables are set:

  • PATH: Should include bin folders of SQLite, Maven, Java, Anaconda
  • JAVA_HOME: Should be set to the main directory (not bin) of your jdk

Setting up the SQLite tables

Navigate to the codegen directory, and run the sh file gen_sqlite.sh

cd codegen ./gen_sqlite.sh cd ..

This will produce the tables necessary to run the modeldb server.

Launching the Server

Now, launch the server using the start_server.sh script. You need Maven installed to do so.

./start_server.sh

Testing the Server

You can now test the server with a sample client by running:

mvn test

The Scala Client

The scala client has a fairly large directory structure. The following commands will assume that you are in the spark.ml directory.

cd client/scala/libs/spark.ml

Environment Variables

You will need the following environment variables set:

  • PATH: Must include the bin directory of sbt (Scala build tool), Anaconda, and Apache Spark.

Assembling the JAR File

Assemble the JAR file using the executable. Internally, this assumes SBT and Anaconda are in your PATH.

From client/scala/libs/spark.ml/

./build_client.sh

This will create a jar:

target/scala-2.11/ml.jar

You can then use this jar in your projects.

Testing the JAR with Sample Projects

ModelDB also includes a few sample projects for you to run. Let's run one to make sure the target compiled correctly.

First, download the adult data set from http://archive.ics.uci.edu/ml/datasets/Adult

Then, from the spark.ml directory, run

spark-submit --master local[*] --class "edu.mit.csail.db.ml.modeldb.sample.CompareModelsSample" target/scala-2.11/ml.jar <path_to_adult.data>