Re: Visit on multiple frinkiac nodes update

Hello,

The issues with Visit on the cluster seem to be resolved now. I tried a
plot running across multiple nodes and it worked without issue.

I know that some of the issues reported involved situations other than
trying to run visit using multiple nodes, but I haven't experienced
anything like that.

So, if you are having these problems, what you should do is make sure
you have the latest basin kernel and basin remote. Make sure you do an
autoreconf for both of these packages. Then the whole "./configure;
make; make install" routine.

I ran my test by starting the cluster using basin_remote from the
command line. Then I ran "ipython long_visit_demo.ipy". This script can
be found in the basin_remote/trunk/scripts directory. If the script is
successful, it should run and then exit. At this point, you can connect
visit by opening the sim file and you should have the following plots
available:

"Density Real" - a continuous plot
"Locations" - a discrete plot

these both come from one of the fake data examples.

There are also plots of var0-var3 which are just some tests from
generated data.

Please let me know if you continue to experience issues or if everything
works fine for you. Once it seems that the issues are worked out, I'll
update frinkiac to visit 1.6.1 (the latest version).

Thanks,
~doug

Doug Jones wrote:
> Hello all,
>
> I've debugged using Visit on multiple frinkiac nodes and it turns out
> that the error is not in the code or the configuration, but a weird
> library resolution issue on nodes other than frinkiac1.
>
> For some reason, the openGL libraries that are installed on the system
> cannot be found on (what I guess to be) nodes > 1.
>
> I'll get in touch with Dan and see if he can figure out what the issue
> is. I believe gentoo caches dynamic library paths, so perhaps these
> became out of date on the children nodes and just need to be flushed and
> updated.
>
> ~doug
>
Received on Tue Oct 2 12:14:44 2007