Capabilities

1. File transfer system: Gridmake is able to transfer local files to remote machines. If you start gridmake in a directory, it transfers all files in that directory including all subdirectories. It is possible to set parameters for files and directories, which should not be transferred. After transferring the whole source code during the first run, gridmake creates a file list and is able to detect changed or new files and transfers only these. Gridmake is able to make use of CVS.
2. Command execution: The primary idea is that gridmake has to be able to start a local make command on target machines, but you can specify any number of commands for gridmake to execute. For the execution of commands the user can tell gridmake to set environment variables on the target machines.

Requirements

• Gridmake is fully implemented in Perl. It requires Perl 5 or higher.
• For communication, gridmake uses ssh. Ssh must be installed and you should generate a private/public key pair for it, if you don't want to type passwords. Another possibility is to use gsi ssh.
• For compression and packaging of the transferred files, gridmake uses tar and gzip. If they aren't installed, gridmake will not work.
• Gridmake is implemented for UNIX and Linux operating systems. A Windows port is not planed. MacOS X is not tested yet.

Installation

Using gridmake

How it works

Gridmake's internal steps

1. Checking commandline options.
2. Reading configfile.
3. a) If this is the first run, gridmake creates a file list of all files in the given directory and all subdirectories, except the files, which you want to be ignored.
   b) If this is not the first run (filelist.gridmake exists), gridmake creates a filelist and compares it with the original list. All new or changed files will be memorized. Removed files will be deleted from the list and not transferred.
4. The files of the final list will be stored in a tar archive and compressed with gzip.
5. The archive will be transferred using the scp command. Gridmake transfers all files simultaneously to all target all machines.
6. Gridmake starts with the client operation. The following steps are performed simultaneously for each target machine as a child process.
7. The gridmake client source code will be generated according to the parameters of the gridmake configfile.
8. The client program will be transferred to the target machine with scp.
9. The gridmake child process starts the client program on the target machine with ssh.
10. The client will perform the following actions, all operations will be logged:
    a) Create target directories if they don't exists.
    b) Move the compressed archive into the directory and extract it.
    c) Execute the list of commands step by step.
    d) If no error occurs, set the return value to zero at the end of the log file, and to one if any errors occur.
11. The log file will be returned by the child process and the return value will be checked.
12. The parent program waits until all child processes finish and displays the result of the gridmake run.

Gridmake with cvs

1. Checking commandline options.
2. Reading configfile.
3. Gridmake starts with the client operation. The following steps are performed for each target machine in a child process.
4. The gridmake client source code will be generated according to the parameters of the gridmake configfile.
5. The client program will be transferred to the target machine with scp.
6. The gridmake child process starts the client on the target machine with ssh.
7. The client will perform the following actions, all operation will be logged:
   a) Create target directories, if they don't exists.
   b) The cvs command will be executed.
   c) Execute the list of commands step by step.
   d) If no error occurs, it sets the return value to zero at the end of the log file, and to one if any errors occur.
8. The log file is returned by the child process and the return value is checked.
9. The parent program waits until all child processes terminate and displays the result of the gridmake run.

Documentation