Start Cavepainting manually in the Cave: Difference between revisions

From VrlWiki
Jump to navigation Jump to search
Jadrian Miles (talk | contribs)
No edit summary
Replaced content with "This page is no longer useful, sorry!"
 
Line 1: Line 1:
[[Cavepainting]] in the [[Cave]] involves coordination among a bunch of different machines.  This article shows you HOWTO get everything running by issuing commands from the command line.
This page is no longer useful, sorry!
 
An important thing to keep in mind when manually troubleshooting CavePainting problems is that the definition of <code>$G</code> is different depending on the machine, user, and current process.  Note especially that the CCV <code>$G</code>, which is what <tt>wallconsole</tt> and <tt>serial</tt> see, has different contents than the CS <code>$G</code>, which <tt>cs-nodes</tt> sees.  When in doubt, check the actual definition of <code>$G</code> in the context that you're investigating.  To get an idea of all the different <code>$G</code>s you might see just on the CCV machines, try running: <pre>
ls /share/cs137 | grep "dollar"
ls /share/gfx | grep "dollar"
ls /share/gfx2 | grep "dollar"
</pre>
 
== VRPN Tracking on <tt>serial</tt> ==
# <code>ssh cavedemo@serial</code>
# <code>cd /share/gfx/bin</code><br>Note that you '''don't''' want to go to $G!
# <code>./vrpn_start.current</code>
# Leave this program running and open up a new terminal to do the following steps.
 
== Event Handling on the Windows Machine ==
# Make sure that <tt>grexecd</tt> is running on the Windows machine in the corner to the right of the Cave and that there are no error messages or other weird things going on there
#* If it's not running, look for the icon on the desktop and double-click it.
 
== Display Control on the clusters ==
 
=== If running from the CCV cluster ===
# <code>ssh -Y front</code> <!--
 
Note that there's no equivalent here to the CS cluster's setup_cave.sh.  The contents of that script are:
 
#!/bin/sh -v
 
nodes="cs-front cs-left cs-right cs-floor cs-top"
config_dir="/vol/projects/cave-powerwall"
 
for x in $nodes; do
  scp $config_dir/cave/$x $x:/etc/X11/xorg.conf
  ssh -x $x /bin/rm -f /home/caveman/session/SESSION
done
 
When switching control of the cave between the CS cluster and the CCV cluster, a script like this is probably important.
 
-->
# <code>/usr/bin/nvidia-settings --ctrl-display=front:0.0</code>
# In the Nvidia config GUI program that pops up:
## Go to "Frame Lock" in the left-hand pane
##* ''If no computer names are displayed in the main pane'', click the "Add Devices..." button and one-by-one add <tt>front:0.0</tt>, <tt>left:0.0</tt>, <tt>right:0.0</tt>, and <tt>floor:0.0</tt>.
## Make sure that <tt>front</tt> has the "server" box checked and all others are "client"
## Hit the "Enable Frame Lock" button in the lower-right
## Once all the red "Timing" and "Stereo" lights along the right turn green, you can hit "Quit"
 
=== If running from the CS cluster ===
# <code>ssh -Y cs-nodes</code>
# <code>cd /vol/projects/cave-powerwall</code>
# <code>./setup_cave.sh</code>
# <code>./myframelock.sh</code>
#*Note that this script might whine.  Give it a long time to throw up errors; it eventually shows.
#*If it stalls after displaying "cs-front", hit ctrl-C and see if it continues through the other machines and works.
#*If it is taking especially long, hit ctrl-C a few times until you return to the prompt and try again.  I'm not sure why this program is so nondeterministic.
# In the Nvidia config GUI program that pops up:
## Go to "Frame Lock" in the left-hand pane
##* ''If no computer names are displayed in the main pane'', click the "Add Devices..." button and one-by-one add <tt>cs-front:0.0</tt>, <tt>cs-left:0.0</tt>, <tt>cs-right:0.0</tt>, and <tt>cs-floor:0.0</tt>.
## Make sure that <tt>cs-front</tt> has the "server" box checked and all others are "client"
## Hit the "Enable Frame Lock" button in the lower-right
## Once all the red "Timing" and "Stereo" lights along the right turn green, you can hit "Quit"
 
== CavePainting Itself on <tt>wallconsole</tt> ==
You have three options, listed in increasing order of hassle and debuggability:
# Press the button on the [http://wallconsole/kiosk/index.cgi?mode=cs137fall08 kiosk webpage] corresponding to your project.
# Run your personalized Cavepainting script directly:
## <code>ssh cavedemo@wallconsole</code>
## This step differs depending on what cluster you want to run CavePainting:
##* ''If running from the CCV cluster'': <code>kiosk/<your tab on the kiosk>/<your script name></code>
##* ''If running from the CS cluster'': <code>$G/bin/cave.cscluster.<some other junk that includes your username></code>
# Run the commands typically included in such a script manually:
## <code>ssh -Y cavedemo@wallconsole</code>
## <code>$G/bin/cave_display_cscluster</code>
## <code>ssh -Y demo@cs-nodes</code>
## <code>setenv GROOT /ccvdisk2/dollar-g-cs137; source $GROOT/tools/gfxtools-startup-csh</code>
## <code>$G/bin/cavepainting-cave -c SaveBasePath=$G/data/cavepaintings/<user> -c LoadBasePath=$G/data/cavepaintings/<user> -c UserDataDir=$G/data/cavepaintings/<user></code>
 
You should now see activity in a new terminal window on the Windows machine, then a big grey window on the Windows machine, and then Cavepainting should start in the Cave.
 
If there are problems with the display, check out [[Troubleshoot Cavepainting|HOWTO Troubleshoot Cavepainting]].
 
Be sure to follow the [[Use Cavepainting|Cavepainting user's guide]], and especially to shut down properly by using the ESC key on the Windows machine.
 
== Manually Hard-killing CavePainting ==
<tt>/users/cavedemo/bin/kill_all_cave_processes</tt>
 
[[Category:CavePainting HOWTO]]

Latest revision as of 15:07, 19 September 2013

This page is no longer useful, sorry!