ipcBeginProcess

[ecnedad]

I encounter a problem in my Remote Simulation.

Using the following setup:
Local Machine: myLocal
Remote Machine: myHost
directory: /home/testfile

/home/testfile was shared to "myHost".

I have used the following commands:
cmd = "rsh myHost; cd /net/myLocal/home/testfile; source /user/
setup/setic51; ./runSpectre"
cid = ipcBeginProcess( cmd "myHost")

Then the simulation finished in konsole in just a couple of seconds.
But even the simulation was done (the spectre finished the simulation
in konsole), the cid is still active.

When I used ipcWaitForProcess(cid), it makes the process longer before
it terminated (ipc:-1).
When I used ipcIsAliveProcess(cid) to test if process is still active,
it took little time compare to ipcWaitForProcess(cid).

Why is that the status of the process was not yet finished even the
simulation is already done and the results file were already
generated?

 

[Andrew Beckett]

ecnedad wrote, on 07/08/08 11:24:

I encounter a problem in my Remote Simulation.

Using the following setup:
Local Machine: myLocal
Remote Machine: myHost
directory: /home/testfile

/home/testfile was shared to "myHost".

I have used the following commands:
cmd = "rsh myHost; cd /net/myLocal/home/testfile; source /user/
setup/setic51; ./runSpectre"
cid = ipcBeginProcess( cmd "myHost")

Then the simulation finished in konsole in just a couple of seconds.
But even the simulation was done (the spectre finished the simulation
in konsole), the cid is still active.

When I used ipcWaitForProcess(cid), it makes the process longer before
it terminated (ipc:-1).
When I used ipcIsAliveProcess(cid) to test if process is still active,
it took little time compare to ipcWaitForProcess(cid).

Why is that the status of the process was not yet finished even the
simulation is already done and the results file were already
generated?

I'm assuming this is a separate question from the follow-on questions in this
thread (which appear to be from someone else).

Without knowing what is in "runSpectre" it's hard to tell. Perhaps spectre
is being launched in "interactive" mode, in which case it doesn't exit at
the end - to allow it to be efficiently re-launched without having to
completely restarted.

By the way, why are you doing a "rsh" in your command as well as specifying
you want to run it on "myHost". In fact, I've just realised what the problem
will be...

Your command is

rsh myHost; cd /net/myLocal/home/testfile; source /user/setup/setic51; ./runSpectre

what that will do is an interactive rsh, starting a shell on the remote machine.
Only when the rsh finishes will it do the cd, source, and runSpectre.

Note that the shell used will probably be Bourne shell too, so the source won't
work.

There have been previous (recent) postings in this forum about sourcing files,
and using csh as part of an ipcBeginProcess. I suggest you google search this
forum for ipcBeginProcess and take a look.

Regards,

Andrew.