PASS: A Java repository from qiuxiangdong

PASS - Programming Assignment Submission System

PASS is an scalable web-based solution for programming assignment submission and automated testing. It is suitable for grading programming assignments in computer science classes wherein student solutions are tested against a set of test cases. It can compile and test C/C++ programs out of the box and it can be extended through bash scripts to support any other programming language.

Dependencies

PASS has the following run-time dependencies:

Java 1.8
MySQL 5.7
GlassFish 4.1
Linux environment
SimpleLinuxSandbox (optional)
RabbitMQ 3.6.6 (optional)
jsvc 1.0.6 (optional)

The sandbox is needed for the C/C++ test script. RabbitMQ and jsvc are only needed if using distributed task scheduler (see the next section).

Components

The system has multiple components:

Database server
File server
Web server
Message broker (optional)
Workers (optional)

Long-running tasks, such as compiling and evaluating submissions, can be handled in two different ways:

Using a local task scheduler: tasks are executed in the web application using a fixed size thread pool

or
Using a distributed task scheduler and one or more workers: tasks are executed by standalone workers using RabbitMQ as the messaging backend

The default is to use workers, you can change that by modifying the following lines in pass.web.common.Container:

    public void initialize(ServletContext context)
    {
        try
        {
            //taskManager = new LocalTaskManager(WebConfig.getInstance().getConcurrentTasks());
            taskManager = new DistributedTaskManager();
            ...

To use the local task scheduler, comment/uncomment the appropriate lines shown above.

The file server stores files related to assignments, student submissions, etc. If using the distributed task scheduler, all workers as well as the web server need to have access to the file server, database server and the message broker. To share files stored on the file server among the web server and workers, you need a distributed file system like NFS. Workers need to have read/write access to these files.

How to Deploy

Step 1: Create the Database

Use the sql script create_tables.sql to generate the MySQL database. Then create MySQL users for the web application and workers as necessary.

Step 2: File Repository

Copy the folder file_repository to your preferred location on your file server.

In order for the students to be able to sign up in the web site, you need to prepare a list of enrolled students as a CSV file with the following format:

student id, username, last name, first name, email

For example:

1234567890,jdoe,Doe,John,jdoe@gmail.com

This file should be stored as list.csv in the file repository.

Step 3: Generate Master Key

You need Maven 3.0+ to build PASS. After building for the first time, you need to run the command line tool located in pass-commandline which has the following usage:

$ java -jar pass-commandline/target/pass-commandline-1.0.jar
Usage: pass-commandline-1.0.jar COMMAND [ARGS]

COMMAND:

    key:    generate master key
    enc:    encrypt a password with master key
    admin:  register admin user in database
    db:
            generate database creation script
            args: <output_file>
    cfg:
            generate all configuration files
            args: <template_dir> <output_dir>
    cfg1:
            generate a single configuration file
            args: <template> <output>

Use the key command to generate a master key. This key will be used for encrypting credentials such as database password in configuration files. You need to keep this key in a safe place.

Step 4: Generate Configuration Files

Use the command line tool with cfg command to generate the configuration files. You need to specify the location of the templates as well as the output directory:

$ java -jar pass-commandline/target/pass-commandline-1.0.jar cfg config_templates/ config/

It will ask you to provide various configuration settings step by step. At some point it will ask you to enter the master key to encrypt passwords. The following tables summarize the required settings:

`basic_settings.xml`
`site_title`	The website title displayed in navbar
`server_url`	The website url, used for referrer validation
`keep_submissions_per_user_project`	How many submissions to keep from each user per assignment

`broker_settings.xml`
`broker_host`	The hostname or IP address of the RabbitMQ server
`broker_user`	The RabbitMQ account username
`broker_password`	The RabbitMQ account password

`database_settings.xml`
`database_url`	The database connection string
`database_username`	The database account username
`database_password`	The database account password

`filesystem_settings.xml`
`repository_path`	The filesystem path for storing submissions and other files
`master_key_path`	Where to find the master key

`web_settings.xml`
`server_email_address`	The email address to use for sending verification codes
`email_server`	The smtp server for `server_email_address`
`email_user`	The username for authenticating to the smtp server
`email_password`	The password for authenticating to the smtp server
`concurrent_tasks`	Number of threads for task execution if using `LocalTaskManager`

Most of these configuration files have two versions: a .web version and a .worker version. For example, you might want to have two different database accounts with different permissions for the web server and the workers. filesystem_settings.xml also has two versions: .local and .remote. The local version is used with the web server and the worker running on the same server, while the remote version is used for workers running on other machines. The rationale is that these servers might be configured to access the files repository differently.

There are other configurations that have default values (not listed here). You can change those settings either in the templates or in the generated configuration files.

Step 5: Create `admin`

After generating the configuration files, re-build PASS. Then use the command line tool to create the admin user in database:

$ java -jar pass-commandline/target/pass-commandline-1.0.jar admin

At this point you can deploy the web application and login with the admin user.

Step 6: Run Workers (optional)

If you decided to use distributed task scheduling, you will need to run one or more workers. Use the provided scripts to run workers as daemons. Make sure to check the worker's log file after starting it. The worker's current log file can be found at /tmp/pass_worker0.log.

Each worker daemon creates two job consumers by default. You can change this setting in env.sh:

ARGS="2 $(hostname)"

A reasonable number would be the number of CPU cores exclusively available to the worker.

Storing the Master Key

Web applications in general are prone to various attacks that could compromise the information stored on the server. To protect database password and other credentials stored in configuration files (encrypted with the master key) in case of such attacks, you need to protect the master key.

Since the web application reads the master key once during its startup, the file containing the master key can be removed safely after startup. This can be achieved for example by providing a readable copy of the master key file only during the startup phase.

qiuxiangdong/PASS