CS176B: Winter '11

Homework 2

The homework is due Friday, February 11th at 11:59pm. All assignments are to be carried out individually. No collaboration is allowed.

The Assignment

In this assignment, you will implement a swarming FTP client application using the socket API.  The application is called mftp, and must be implemented in C.  The FTP protocol is specified in RFC 959.  The assignment can be broken down into two components: basic ftp operation and swarming downloads.  Both components must be completed by the due date.

Basic Operation

The client application will take an FTP server and a file name as command-line parameters. When executed, the client will simply retrieve the file specified using the FTP protocol and save it to a local file with the same name. Note: you will not need to implement the put operation.

For example, by executing:

% mftp -s ftp.ucsb.edu -f /robots.txt

The file named robot.txt will be created in the current directory.

On success, the mftp application exits with code 0. If an error occurs, the application exits with the the appropriate code and prints a human-readable error message on standard error.

Other command line options include:

-h or --help

Prints a synopsis of the application usage and exits with return code 0.

-v or --version

Prints the name of the application, the version number (in this case the version has to be 0.1), the author, and exits, returning 0.

[-f file] or [--file file]

Specifies the file to download.

[-s hostname] or [--server hostname]

Specifies the server to download the file from.

[-p port] or [--port port]

Specifies the port to be used when contacting the server. (default value: 21).

[-n user] or [--username user]

Uses the username user when logging into the FTP server (default value: anonymous).

[-P password] or [--password password]

Uses the password password when logging into the FTP server (default value: user@localhost.localnet).

-a or --active

Forces active behavior (the server opens the data connection to the client) (default behavior: passive behavior).

[-m mode] or [--mode mode]

Specifies the mode to be used for the transfer (ASCII or binary) (default value: binary).

[-l logfile] or [--log logfile]

Logs all the FTP commands exchanged with the server and the corresponding replies to file logfile. If the filename is "-" then the commands are printed to the standard output. The lines must be prepended by C->S: or S->C: for client-to-server and server-to-client lines, respectively. For example:

S->C: 220 ProFTPD 1.2.10 Server (UCSB) [128.111.24.43]
C->S: USER anonymous
S->C: 331 Anonymous login ok, send your complete email address as your password.
C->S: PASS user@localhost.localnet

Multi-server Swarming

The next part of the assignment is to implement a "swarming" version of ftp download.  The idea is that the mftp client uses separate threads to simultaneously connect to multiple servers, and download the same file from different positions in the same file.  On the server side, this functionality is already supported by current ftp servers through the "restart marker" option.

There's no intuitive user interface for this type of functionality. So we will use something very simple and straightforward.  mftp takes in an additional commandline argument:
    [-w swarm-config-file] or [--swarm swarm-config-file]
    Specifies the login, password, hostname and absolute path to the file

When mftp reads in the swamp-config file, it parses the N servers on the list.  For each of the N servers, mftp spawns a separate thread, connects to the server, and moves to the specified directory.  To download a file of S bytes, mftp performs a swaming download, where each thread downloads a segment of S/N bytes. The downloaded segments should be put together to form the entire file.  Note: mftp can use the --swarm option simultaneously with the -l option.

The format should be: ftp://username:password@servername/file-path. An example of the swarm-configuration file is as follows:  

Discussion

This application must be extremely resilient to wrong/unexpected input data, repeated values, etc. Part of your assignment is to understand how the mftp application could be abused by a user and to provide meaningful error messages if something goes wrong. You have to program defensively and with user-friendliness in mind. Segmentation fault is not an option!

You CANNOT use any existing library to retrieve the files from the server. You have to write your own code that will open connections, play the protocol, parse the data received from the server, etc.

The application should return meaningful error messages if an error occurs. In addition, the exit value should give some information about what happened. By returning different exit values in association with different errors the application becomes easily scriptable (that is, an external script can invoke the application and manage error conditions).

Obviously it is impossible to foresee all possible error conditions, therefore the following list is obviously incomplete.

Exit code	Explanation
0	Operation successfully completed
1	Can't connect to server
2	Authentication failed
3	File not found
4	Syntax error in client request
5	Command not implemented by server
6	Operation not allowed by server
7	Generic error

Miscellaneous notes

• Note that when invoked without any parameters the application should behave as if it were invoked with the --help option.
• All error message must be printed to standard error.
• If a parameter is specified multiple times, then the last value assigned to the parameter is the value accepted.

Submission

The submission process for this assignment uses the turnin package. Do a man turnin to find more info about this program. The name of the assignment is mftp.

Your submission must contain one source file called mftp.c that contains all code for your application. In addition, you have to write a brief README file that describes your application (and possible issues and problems). You also have to include a file called AUTHORS that contains your name and your email address. Finally, you need to include a Makefile file that describes how to build your application.

To submit:

1. Create a directory whose name is your CS account. For example, user John Doe --whose account is jdoe-- would do:

   % mkdir jdoe
   

2. Put in the directory all  files for homework 2.
3. Execute the turnin program. For example, user jdoe would execute:

   % turnin mftp@cs176b jdoe
   

You can execute turnin up to 10 times per project. Earlier versions will be discarded. The timestamp of turnin has to be before the due date.