Jean,
I also found this bug in the last few days.  It is written up as bug 14732.  
Since your description below is quite good, I added it to my bug report as a 
note.  And yes, this is a ParaView bug.

The only way I have found to get around this issue is to delete the .ini file 
before running ParaView.  ParaView will just create a new, clean .ini file that 
works fine.  But, you have to delete the file every time - the next time you 
run will be bad again.  I delete this file in a script that users run, which 
wraps the real call to ParaView.

I believe that the root cause of the issue is that you (and I) have different 
plugins compiled on your client and clusters, and possibly between clusters.  
Thus, when you try to connect to a cluster, ParaView is looking for a plugin 
remotely that does not exist.  That's my theory, anyway.

Alan



-----Original Message-----
From: ParaView [mailto:[email protected]] On Behalf Of Favre Jean
Sent: Thursday, May 15, 2014 3:08 AM
To: [email protected]
Subject: [EXTERNAL] [Paraview] Manage Plugin issues with auto-load

hello

I have multiple production machines where PV is installed, some with GPUs, some 
with mesa-rendering. Even so, operating systems differ on the different 
machines.

The problem I have is that PV is installed in application directories on a 
filesystem shared by all production machines. And when using the auto-load 
feature of Manage Plugin, I often have conflicts, or even crashes when 
establishing a connection. For example:

Wrong tag but don't know how to handle it... 188971 Aborted (core dumped)

I use reverse-connection to connect to all production machines. My servers.pvsc 
file has multiple entries, each one looking like

<Server name="Reverse-Connect-Cluster1" resource="csrc://"> ...
</Server>
<Server name="Reverse-Connect-Cluster2" resource="csrc://"> ...
</Server>
<Server name="Reverse-Connect-Cluster3" resource="csrc://"> ...
</Server>

what I am finding out is that my desktop init file "ParaView4.1.0.ini" 
remembers which plugins to auto-load based on a tag composed of "csrc://port"

thus, when using the default port 11111 for all connections, I am loading the 
plugins referred to under "csrc://11111" regardless of the cluster's name. This 
causes conflicts or crashes on connecting.

seems like a fix for a single user would be to always use different port 
numbers for each production machine. But then I don't know how to generalize 
this for all users in the Computing Center.

Any tip on how to avoid conflicts with port numbers and reverse connection, and 
ensuring that auto-load works properly would be welcome.

TIA

-----------------
Jean

_______________________________________________
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the ParaView Wiki at: 
http://paraview.org/Wiki/ParaView

Follow this link to subscribe/unsubscribe:
http://www.paraview.org/mailman/listinfo/paraview
_______________________________________________
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the ParaView Wiki at: 
http://paraview.org/Wiki/ParaView

Follow this link to subscribe/unsubscribe:
http://www.paraview.org/mailman/listinfo/paraview

Reply via email to