Hi, Star users?

Have you guys experienced that STARWATCH DAEMON is prone to die?

I am suffering a lot from this. The daemon (starwdaemon) is not always responsive to the clients (starwatch). If I got such a warning message

"Warning - Server response timed out. Star job unable to check for registeration with starwatch."

I then had to kill starwatch and consequently starwdaemon was also dead. Another way is that I kill directly starwdaemon, so starwatch is alive but becomes useless. In either way I lost monitoring of some other running jobs which may have good communications with the daemon.

For my own experience, as long as the job size is of about 50MB or above, starwatch/starwdaemon becomes practically useless.

Any idea or comments?

Thanks,

Jiaying

http://www.beilke-cfd.de/plot_runfile.pl

Hello Joern,

could you please give a short explanation or documentation of this script, and how it has to be used?

You need perl and gnuplot to use the script.

1) make it an executable : chmod 755 plot_runfile.pl

2) run it from the directory where you have the runfile by entering: plot_runfile.pl casename

The script takes the runfile (case.run) removes the header lines and put additional spaces between the rows so that gnuplot can work with it. Then gnuplot is started and the residuals and the monitoring values are plotted.

Works perfectly!

Thanks Joern!

Jiaying

It's a nice script, but I'm solving for an extra scalar and unfortunately, the extra line produced in the .run file for each iteration makes the plots appear rather ugly.

Regards,

Lars Ola

There is a slightly improved version now on the server.

It still can't handle the extra line per iteration in the run file when you solve for additional scalars. Otherwise it works, and looks, perfectly.

Regs.

Lars Ola

If you are willing to pay for that you can get it

hi usres, how can one add extra variable like mass flow rate in *.run file??

nikhil

hi users, how can one add extra variable like mass flow rate in *.run file??

nikhil

Use the posdat.f subroutine and write to unit 60 whatever you want.

Or hava a look to the star-guide

Analysis Montrols -> Output Controls -> Monitoring ... The selected data are written to the *.erd file.

Thanx Joern, the seecond way that u told i am using this time, i want it be in *.run file so that i can use ur script directly to see the profile of any other variable (not only mass flow rate) during code is running.

nikhil

hi joern, there is one drawback of the script ?suppose the number of iteration is greater than 10000 then there is overlapping of plots of (ite no 1 to ite no 10000 and beyond ite no 10001) bcoz in *.run file it shows iteratons beyond 10000 like iteration 1 .

can this bug be modified? nikhil

Hangs up often in Windows, but I've not noticed any correlation between job size and starwatch daemon stability -I've done an underhood sim overnight (under Windows) and by next morning starwatch is still dutifully reporting all the residuals. I have never gotten it to run at all in Linux -the daemon and the stripchart were trying to reach each other on wrong ports! However, starwatch is very stable on UNIX platforms.

Now, I mostly just stare at the text residual outputs a'la 'The Matrix'.