If you see the message "Xlib: client is not authorized to connect to Server" then
you may have forgotten to allow the remote machine running enfuzion to
open windows on your machine. To remedy this, give the command
xhost +<remote_machine_name>
at your local machine. To run enfuzion on hathor this will be
xhost +hathor.csse.monash.edu.au
You may also need to tell the remote machine where to display its output by
setting the DISPLAY variable. On your remote machine give the command
setenv DISPLAY <local_machine_name>:0
The actual values to be used are decided in the Generator phase, where the user has a chance to select from the "values". If this is not done then the Dispatcher will use the values specified as "defaults".
You may have neglected to create a node configuration file on the remote machine.
Or perhaps it is not on the path there. The file must be called enfuzion.nodes and
has a format of the following form:
| <first_machine_name> | <your_user_name_for_that_machine> | <password_for_that_machine (optional)>
| <second_machine_name> | <your_user_name_for_that_machine>
| <password_for_that_machine (optional)>
| <third_machine_name> | <your_user_name_for_that_machine>
| <password_for_that_machine (optional)>
| ... | ... | ...
| |
If all the machines have the same password then there is a simple solution.
As an example take the enfuzion network at hathor (the PPME) running under linux. The nodes are named
cluster01, cluster02, ..., cluster26.
Suppose that your login name is fred and your password is newuser (shame) on
all these node. Then the following enfuzion.nodes file will prompt only for the
initial password.
|
You may have clicked the generator button on the front menu. Instead you should have pulled down the FILE menu in the preparator window and selected the generator item there. Similarly, when the generator window is finalized pull down the FILE menu to select the Dispatch option to continue to the final phase.
The error messages should indicate which command is failing and on which nodes.
Ensure that this command will execute on one of those nodes by logging in and
by running the command there from the command line. The command may fail
because it uses an application that is not available in your hathor account,
or it needs some input files that are not there.
There may also be incompatabilities with the operating system. For example,
if the application is compiled under Unix System V, then it may not run under
the linux operating system on hathor.
If the command executes correctly when manually entered in the root node but not
within enfuzion then the problem may be caused by a path that does not encompass
the files used. Note that when running enfuzion the working directory will not be
your home directory on the server. The solution is to adjust your path or to use
absolute addressing for the file names. So user
fred operating under linux, running
an application fredsthing in her home
directory, might issue the command
No. If you retrace your steps (via the FILE menu) to the generation screen,
then the parameters shown there can be manually adjusted. If you further
backtrack to the preparator phase, then the plain text copy of the plan
shown may be manually edited as a text document.
At the front end, click the appropriate button for the phase that you require:
Preparator, Generator or Dipatcher. Under the FILE menu pull down "Open ..."
and then enter the full name (including extension) of the plan required.
In the dispatch phase, some jobs failed, generating many
error messages. How to proceed?
~/fredsthing > output
If I have just made an enfuzion run and now want to modify
the commands and/or the parameters. Do I need to start again with a new plan?
When I enter the enfuzion front end how do I summon up an
old plan?