allankv / griffin

Model driven data quality service

Griffin

Griffin is a Data Quality solution for distributed data systems at any scale in both streaming or batch data context. It provides a framework process for defining data quality model, executing data quality measurement, automating data profiling and validation, as well as a unified data quality visualization across multiple data systems. You can access our home page here.

Contact us

Google Groups

CI

https://travis-ci.org/eBay/griffin

Repository

Snapshot: https://oss.sonatype.org/content/repositories/snapshots

Release: https://oss.sonatype.org/service/local/staging/deploy/maven2

How to build

1. git clone the repository of https://github.com/eBay/griffin
2. run "mvn install"

How to run in docker

1. Install docker.

2. Download our docker folder to your work path.

3. Enter docker directory and build images.
   The first step is to build griffin-base-env, which prepares the environment for griffin.

   cd <your work path>/docker/griffin-base
   docker build -t griffin-base-env .
   

   The second step is to build griffin-env, which contains examples for griffin demo.

   cd <your work path>/docker/griffin
   docker build --no-cache -t griffin-env .
   

4. Run docker image griffin-env, then the backend is ready.

   docker run -it -h sandbox --name griffin -m 8G --memory-swap -1 \
   -p 2122:2122 -p 47077:7077 -p 46066:6066 -p 48088:8088 -p 48040:8040 \
   -p 48042:8042 -p 48080:8080 -p 47017:27017 griffin-env bash
   

   You can also drop the tail "bash" of the command above, then you will get tomcat service log printing in docker only.

5. Now you can visit UI through your browser, and follow the next steps on web UI here. You can login with account "test" and password "test" if required.

   http://<your local IP address>:48080/
   

   And you can also ssh to the docker container using account "griffin" with password "griffin".

   ssh griffin@<your local IP address> -p 2122
   

How to deploy and run at local

1. Install jdk (1.7 or later versions)

2. Install Tomcat (7.0 or later versions)

3. Install MongoDB and import the collections

   mongorestore /db:unitdb0 /dir:<dir of griffin-doc>/db/unitdb0
   

4. Install Hadoop (2.6.0 or later), you can get some help here.
   Make sure you have the permission to use command "hadoop".
   Create an empty directory in hdfs as your hdfs path, and then create running and history directory in it

   hadoop fs -mkdir <your hdfs path>
   hadoop fs -mkdir <your hdfs path>/running
   hadoop fs -mkdir <your hdfs path>/history
   

5. Install Spark (version 1.6.x, griffin does not support 2.0.x at current), if you want to install Pseudo Distributed/Single Node Cluster, you can get some help here.
   Make sure you have the permission to use command "spark-shell".

6. Install Hive (version 1.2.1 or later), you can get some help here.
   Make sure you have the permission to use command "hive".

7. Create a working directory, and it will be your local path now.

8. In your local path, put your data into Hive.
   First, you need to create some directories in hdfs.

   hadoop fs -mkdir /tmp
   hadoop fs -mkdir /user/hive/warehouse
   hadoop fs -chmod g+w /tmp
   hadoop fs -chmod g+w /user/hive/warehouse
   

   Then, run the following command in your local path

   schematool -dbType derby -initSchema
   

   Now you can put your data into Hive by running "hive" here. You can get sample data here, then put into hive as following commands

   CREATE TABLE users_info_src (
     user_id bigint,
     first_name string,
     last_name string,
     address string,
     email string,
     phone string,
     post_code string)
   ROW FORMAT DELIMITED
   FIELDS TERMINATED BY '|'
   STORED AS TEXTFILE;
   
   LOAD DATA LOCAL INPATH '<your data path>/users_info_src.dat' OVERWRITE INTO TABLE users_info_src;
   
   CREATE TABLE users_info_target (
         user_id bigint,
         first_name string,
         last_name string,
         address string,
         email string,
         phone string,
         post_code string)
       ROW FORMAT DELIMITED
       FIELDS TERMINATED BY '|'
       STORED AS TEXTFILE;
   
   LOAD DATA LOCAL INPATH '<your data path>/users_info_target.dat' OVERWRITE INTO TABLE users_info_target;
   

   If you use hive command mode to input data, remember to create _SUCCESS file in hdfs table path as following

   hadoop fs -touchz /user/hive/warehouse/users_info_src/_SUCCESS
   hadoop fs -touchz /user/hive/warehouse/users_info_target/_SUCCESS
   

9. You can create your own model by modifying code.
   (If you want to use our default models, please skip this step)

10. Currently we need to run the jobs automatically by script files, you need to set your own parameters in the script files and run it.
    You can edit the demo script files as following.

    env.sh

    HDFS_WORKDIR=<your hdfs path>/running
    

    griffin_jobs.sh

    spark-submit --class org.apache.griffin.accuracy.Accu --master yarn-client --queue default --executor-memory 1g --num-executors 4 $GRIFFIN_HOME/griffin-models.jar  $lv1dir/cmd.txt $lv1dir/ >> $logfile 2>&1
    spark-submit --class org.apache.griffin.validility.Vali --master yarn-client --queue default --executor-memory 1g --num-executors 4 $GRIFFIN_HOME/griffin-models.jar  $lv1dir/cmd.txt $lv1dir/ >> $logfile 2>&1
    

    These commands submit the jobs to spark, if you want to try your own model or modify some parameters, please edit it.
    If you want to use your own model, change "$GRIFFIN_HOME/griffin-models.jar" to "your path/your model.jar", and change the class name.

    Put these script files in your local path.

11. Open application.properties file, read the comments and specify the properties correctly. Or you can edit it as following.

    env=prod
    job.local.folder=<your local path>/tmp
    job.hdfs.folder=<your hdfs path>
    job.hdfs.runningfoldername=running
    job.hdfs.historyfoldername=history
    

    If you set the properties as above, you need to make sure the directory "tmp" exists in your local path

12. Build the whole project and deploy.

    mvn install -DskipTests
    

    Find the griffin-models-0.1.0-SNAPSHOT.jar in path griffin-models/target, rename it to griffin-models.jar (or your model.jar), and put in your local path.
    Till now, please make sure that, in your local path, there exists griffin-modes.jar, env.sh, griffin_jobs.sh and griffin_regular_run.sh.
    Run griffin_regular_run.sh as following.

    nohup ./griffin_regular_run.sh &
    

    Deploy griffin-core/target/ROOT.war to tomcat, start tomcat server, then you can follow the web UI steps here.

13. You can also review the RESTful APIs through http://localhost:8080/api/v1/application.wadl

How to develop

In dev environment, you can run backend REST service and frontend UI seperately. The majority of the backend code logics are in the griffin-core project. So, to start backend, please import maven project Griffin into eclipse, right click griffin-core->Run As->Run On Server

To start frontend, please follow up the below steps.

1. Open griffin-ui/js/services/services.js file

2. Specify BACKEND_SERVER to your real backend server address, below is an example

   var BACKEND_SERVER = 'http://localhost:8080'; //dev env
   //var BACKEND_SERVER = 'http://localhost:8080/ROOT'; //dev env
   

3. Open a command line, run the below commands in root directory of griffin-ui

   • npm install
   • bower install
   • npm start

4. Then the UI will be opened in browser automatically, please follow the User Guide, enjoy your journey!

Note: The front-end UI is still under development, you can only access some basic features currently.

Contributing

See CONTRIBUTING.md for details on how to contribute code, documentation, etc.