jagr.conf

# The locations of the following directories may be configured here
dir {
    # Grader jar ingest directory
    graders=graders
    # Runtime dependencies for submissions
    libs=libs
    # Rubrics export directory
    rubrics=rubrics
    # Submissions ingest directory
    submissions=submissions
    # Submission export directory
    submissions-export=submissions-export
}
executor {
    # 
    # The maximum amount of concurrency to use for grading.
    # For a given concurrency n, Jagr will ensure that a maximum of n threads or processes are used concurrently that actively run
    # submission code.
    concurrency=4
    # 
    # The JVM arguments to use for grading. These arguments are passed to the JVM that runs the grading code.
    # This only applies to the "process" mode, as the "thread" and "single" modes do not spawn a new JVM.
    #        
    jvm-args=[]
    # 
    # The executor mode to use. The following options are available:
    # - "single" ::
    #  Runs every TestCycle consecutively in the main thread. This mode does not create any extra processes or threads for grading.
    # 
    # - "thread" ::
    #  Creates a separate thread for every TestCycle. This mode greatly speeds up the grading process, especially with a large
    #  amount of submissions. The overhead of creating, managing and synchronizing threads is minimal compared to the performance
    #  benefits. However, this mode has the danger of creating "unkillable" threads (e.g. from certain kinds of infinite loops)
    #  which dramatically slow down the grading process through resource starvation of the host machine.
    # 
    #  The maximum number of concurrent threads used for grading is defined by the option "concurrency".
    # 
    # - "process" ::
    #  Creates a separate process for every TestCycle. This mode has the most overhead, but is also the most defensive against
    #  "badly behaving" code. A certain amount of sandboxing can be achieved in this mode, which is not possible in the other modes
    #  such as "thread" or "single".
    # 
    #  The maximum number of concurrent child process used for grading is defined by the option "concurrency".
    mode=process
    # 
    # The grading thread's maximum permitted elapsed userTime in milliseconds since the last timeout before an
    # AssertionFailedError is thrown. If a thread's userTime satisfies
    # (userTime - lastTimeout) > individualTimeout,
    # the current userTime is stored for comparison later, and an AssertionFailedError is thrown to be caught by JUnit.
    timeout-individual=10000
    # 
    # The grading thread's maximum permitted elapsed userTime in milliseconds (from thread start) before an
    # AssertionFailedError is thrown. If a thread's userTime satisfies
    # ((userTime - lastTimeout) > individualTimeout) && (userTime > totalTimeout),
    # an AssertionFailedError is thrown to be caught by JUnit. Note that lastTimeout is not reset in this case, and all further
    # invocations of checkTimeout() will result in an AssertionFailedError
    timeout-total=150000
}
extras {
    moodle-unpack {
        assignment-id-regex=".*Abgabe[^0-9]*(?<assignmentId>[0-9]{1,2}).*[.]zip"
        enabled=true
        student-regex=".* - (?<studentId>([a-z]{2}[0-9]{2}[a-z]{4})|([a-z]+_[a-z]+))/submissions/.*[.]jar"
    }
}
transformers {
    timeout {
        enabled=true
    }
}