Skip to content
This repository has been archived by the owner on May 10, 2023. It is now read-only.

Latest commit

 

History

History
338 lines (260 loc) · 14.1 KB

README.md

File metadata and controls

338 lines (260 loc) · 14.1 KB

Model Management System

Language Grade: Java CircleCI

The MMS AMP is a hosted application run atop the community version of an Alfresco Enterprise Content Management Server.

NOTE 1: AMPs for Alfresco as built from this maven project are meant to be run by Alfresco Community Edition v5.1.g (AKA 201605-GA)

Per Alfresco's documentation on Modules, Alfresco Module Packages (AMPs) are installed using the Module Management Tool MMT jar file. The latest version of the two AMPs you will be "exploding" into the "alfresco" and "share" WARs resident within the embedded Tomcat server are available via the download links leading to the two interdependent portions named "mms-repo.amp" and "mms-share.amp", relating to "alfresco.war" and "share.war", respectively.

Download of the alfresco WAR related portion of the "MMS" Alfresco module.

Download of the share WAR related portion of the "MMS" Alfresco Module.

In tandem with the EMS-Webapp (colloquially known as the View Editor), and the Magicdraw Development Kit (MDK) for MagicDraw client and Teamwork Cloud users; this github repo serves as a one-stop shop to set up the Model Management Server per the MMS-MDK-VE Compatibility Matrix.

Developer Setup

Dependencies

  • ElasticSearch 5.x (Up to 5.5)
  • PostgreSQL 9.x (Up to 9.4 is using PostgreSQL for Alfresco)

Optional Dependencies (optional if using MDK 3.3.3+, only needed for VE realtime updates)

  • ActiveMQ 5.X

1a. Using Intellij

  • Open Project with root of 'mms'
  • Import Maven Project
  • Open Project Structure and Import Module "mms-ent" and set to find projects recursively

1b. Import Project from git repo to Eclipse

  • Eclipse > File > Import > General > Existing Projects into Workspace
  • Set 'Root Directory' as the path to mms e.g. Browse to $HOME/workspace/mms
  • In the 'Projects' list, you should see all poms. Click Finish

1c. Import Maven Project into Eclipse

  • Eclipse > File > Import > Maven > Existing Maven Projects
  • Set 'Root Directory' as the path to mms e.g. Browse to $HOME/workspace/mms/mms-ent
  • In the 'Projects' list, you should see all poms. Click Finish

2. Configure Eclipse to use Maven 3.X.X

  • Eclipse > Window > Preferences > Maven > Installation
  • Toggle Maven 3.X.X
    • If Maven 3.X.X is not listed, download and install it.
    • On a Mac, install it at /usr/local/Cellar/maven.
    • On a Linux, anywhere in your $PATH.
  • Return to Eclipse > Window > Preferences > Maven > Installation
  • Choose Add...
    • Browse and select Maven 3.X.X installed location.
    • Location is the maven home that you can get by running the newly installed maven, with mvn -V

3. Configure Run Configuration**

  • Select mms-ent project
  • From menu bar, choose Run > Run Configurations
  • Right-click Maven Build > New
    • Enter mms for Name textbox
    • At Main tab
      • Enter ${project_loc} or ${workspace_loc} for Base Directory textbox
      • Enter install for Goals textbox
      • Enter run for Profiles textbox
      • Select Maven 3.X.X (whatever you chose to setup in step 2) for Maven Runtime
    • At JRE tab
      • Select Java 8 for JRE.
      • If it's not installed, download and install Java 8. Afterward, return to here and select Java 8.

Install Dependencies

1. Install and Configure ElasticSearch

  • Download ElasticSearch 5.X
  • Install ElasticSearch
  • Start ElasticSearch then run mms-ent/repo-amp/src/main/resources/mms-mappings.sh

2. Install and Configure PostgreSQL

  • Download PostgreSQL 9.x
    • If using PostgreSQL as the database for Alfresco, PostgreSQL 9.4 is the latest supported version
  • Install PostgreSQL
  • Start PostgreSQL server
  • Connect to the PostgreSQL server and:
    • Create a mms user (referenced by pg.user in your mms-ent/mms.properties file)
      • Ensure you set a password (referenced by pg.pass)
    • Create a mms database ( referenced by pg.name)
  • Execute mms-ent/repo-amp/src/main/resources/mms.sql
    • windows CMD e.g.: psql -h localhost -p 5432 -U mms -d mms -v schema=public < C:\path\to\mms\repo\mms.sql

3. Install and Configure ActiveMQ

  • Download ActiveMQ 5.X
  • Install ActiveMQ
  • Start ActiveMQ service

Running

1a. Running Alfresco

  1. Select file menu Run > Run Configurations
  2. Expand Maven Build
  3. Select mms
    1. Click Run button
    • If you get error:-Dmaven.multiModuleProjectDirectory system property is not set. Check $M2_HOME environment variable and mvn script match. Goto Window -> Preference -> Java -> Installed JREs -> Edit -> Default VM arguments set -Dmaven.multiModuleProjectDirectory=$M2_HOME

1b. Running Alfresco

  1. From mms-ent directory, either run clean-run.sh, run.sh, or ./mvnw install -Prun -Ddependency.surf.version=6.3

2. Testing Alfresco

  1. Enter http://localhost:8080/share/ at a browser's url address textbox.
  2. Enter admin for user name
  3. Enter admin for password

Design Documentation

1. MMS using ElasticSearch and PostgreSQL


General Design

+----------------+   \
| REST API Layer |    \
|----------------|     \
|   WebScripts   |      MMS
|----------------|     /
|Storage | Layer |    /
+----------------+   /
  /\         /\
  ||         ||
  \/         \/
+-----+    +----+
| RDB |    | ES |
+-----+    +----+
(Graph)    (Data)

2. Configuration Schema

  • Global PG database called 'mms' holds configuration information per project
  • Contains Org, Project, and DB Location information

3. Graph Database

  • Each project has it's own database configured in the mms database

  • All graph related information stored in relational database

  • Schema defined in mms.sql

    • Nodes
    • Edges
    • Edge types
    • Node types
    • Commits
    • Refs
  • Functions defined that are recursive for getting parents, children etc.

  • All access to the graph is done via the PostgresHelper that contains functions to interface with the database

  • The graph should never be manipulated directly from anywhere else

The nodes in the graph contain pointers to ElasticSearch documents which contain the real information * Same goes for other things such as configurations or commits

Each of these pointers is the “latest one” * Because for a given SYSMLID there can be multiple documents for each version of that node

The history of each node can be retrieved via a query to ElasticSearch (see ElasticHelper.java)

4. Graph Schema

Nodes and edges both have a type associated with them The type must exist in the NodeTypes and EdgeTypes tables

  • Correspondingly, the types also exist in the PostgresHelper code
  • Each node can be assigned a particular type, by default it is just element
  • Each edge can be assigned a particular type, containment being the basic type

Some assertions about the graph:

  • It is always the case that each node in the graph has a single containment parent
    • If you have multiple, something wrong happened!
  • Multiple root parents are only possible for not containment type edges
  • There should never be any orphan nodes in the graph
    • Always have either children or parents
  • All elastic references in the graph must exist in ES

5. Graph

Structure of the graph for each "project"

           +---------+
        +--| Commits |
+===+   |  +---------+          +-------------+    +----------+
| P |   |                   +---| Holding bin |----| Elements |
| R |   |  +-------+        |   +-------------+    +----------+
| O |   +--| Nodes |--------|
| J |   |  +-------+        |   +----------+
| E |   |                   +---| Elements |
| C | --|  +-------+            +----------+
| T |   +--| Edges |
|   |   |  +-------+
|   |   |
|   |   |  +----------------+
+===+   +--| Refs           |
           +----------------+

6. Example webscript in new world: modelpost

Create instance of EmsNodeUtil Get all other relevant information, validation, etc. form request Figure out if the elements are to go to the holding bin or not Calculate the qualified name and ID information for each node Add metadata for each element Store elements in ES Update graph Create JSON response and return

7. Example webscript in new world: modelget

Create instance of EmsNodeUtil Get all other relevant information, validation, etc. form request Access PG to get the ElasticSearch IDs for all documents that we are interested in Access ES via ElasticHelper and get those IDs Create JSON response and return

8. General Pattern for WebScripts

Create instance of EmsNodeUtil Get all other relevant information, validation, etc. form request Understand how to:

  • Add information to ES
  • Update the graph with the corresponding information from ES

All the work is to be done here for each webscript. The pre and post should always be the same.

Create JSON response and return

Testing

Initializing Organizations, Projects, and test elements

Use the following curl commands to post an initial organization + project:

curl -w "\n%{http_code}\n" -H "Content-Type: application/json" -u admin:admin --data '{"orgs": [{"id": "vetest", "name": "vetest"}]}' -X POST "http://localhost:8080/alfresco/service/orgs"
curl -w "\n%{http_code}\n" -H "Content-Type: application/json" -u admin:admin --data '{"projects": [{"id": "123456","name": "vetest","type": "Project"}]}' -X POST "http://localhost:8080/alfresco/service/orgs/vetest/projects"

Then you can post some elements. For convenience, there is a json file in runner/src/test/robotframework/JsonData. Using the project from above:

curl -w "\n%{http_code}\n" -H "Content-Type: application/json" -u admin:admin --data @JsonData/PostNewElements.json -X POST "http://localhost:8080/alfresco/service/projects/123456/refs/master/elements"

Make sure the elements went in:

curl -w "\n%{http_code}\n" -H "Content-Type: application/json" -u admin:admin -X GET "http://localhost:8080/alfresco/service/projects/123456/refs/master/elements/123456?depth=-1"

Robotframework test suite

Robot tests can be run with the following maven profiles in the mms-ent directory:

./mvnw install -Prun,robot-tests

Please note that tests should be run on a clean instance, therefore, it may be helpful to run clean.sh before running the tests

The Robotframework tests require the 'requests' and 'robotframework-requests' python modules. Install it as follows:

pip install --target=runner/src/test/robotframework/libraries requests
pip install --target=runner/src/test/robotframework/libraries robotframework-requests

OR:

pip install --target=$HOME/.m2/repository/org/robotframework/robotframework/{ROBOTPLUGINVERSION}/Lib requests
pip install --target=$HOME/.m2/repository/org/robotframework/robotframework/{ROBOTPLUGINVERSION}/Lib robotframework-requests

Changing debug levels on the fly

If you need to change debug levels on the fly, use the following endpoint.

alfresco/service/loglevel

It takes as input JSON that specifies the classes and the log levels. For example:

[
  {
    "classname": "gov.nasa.jpl.view_repo.webscripts.ModelGet",
    "loglevel": "DEBUG"
  }
]

API Documentation

API Documentation is located at the following endpoints:

Swagger CodeGen:

alfresco/mms/index.html

Swagger UI:

alfresco/mms/swagger-ui/index.html

Swagger YAML file:

alfresco/mms/mms.swagger.yaml

Migrations after 3.2

For versions after 3.2, most notably 3.3.0, an automigration step has been included to run necessary migrations automatically during the initial startup of alfresco after the upgrade. Should this migration fail for any reason, you can trigger the migration manually by accessing the following endpoint:

alfresco/service/migrate/{targetVersion}

where {targetVersion} is the version that you are upgrading to. Example:

alfresco/service/migrate/3.3.0

This operation is idempotent and can be safely run multiple times.

Deployment Requirements

For deploying MMS and VE five machines are recommended + 1 optional TeamWorkCloud machine.

The given configuration below are in terms of AWS instances which define required CPU and RAM, assuming CentOS as operating system.

Data is stored using Amazon EBS volumes, where 100GB is a good start.

For Postgres Amazon's RDS service is used, where 10-15 GB is a good start.

The minimal configuration (can handle about 300k elements) is

  • 1x t3.large for MMS (8GB)
  • 1x t3.large for RDS postgres service (8GB)
  • 3x t3.medium for Elastic cluster (4GB), t3.large for better performance.
  • 1x AWS elb service - load balancer for elastic cluster
  • EBS storage for data (100 GB)

For a single machine configuration you need one which is powerful enough to cover the above.

A 64 GB configuration can hold about 2 billion elements but might be slow, so you might want to switch to a powerful CPU configuration:

t3.xlarge -> r5.large -> r5.xlarge

Contributing

To learn how you can get involved in a variety of ways, please see Contibuting to OpenMBEE.