[Yankee]

Originally Posted by ginggs

New flexlm tools are compiled for Linux Standard Base, if you don't have the lsb-core package installed you will see this error.

To fix:
sudo apt-get install lsb-core

Thank you ginggs, in our network it makes anslic_admin work well on all the computers, including the ubuntu ones.


As a matter of fact, in my network another license-related problem appeared because I had 64bits machines using ansys mounted from a 32bit file server. When I installed ansys 64bits on the 32bit file server, everything seemed to work well (with a warning that the 64bit version is not suitable for the current 32bit machine), but the installer forgot to copy the files from
ansys_inc/shared_files/licensing/linx64/update
to
ansys_inc/shared_files/licensing/linx64
(those files are essential license utilities such as ansysli_client, in their 64bit version)

That made workbench unable to get a license (whereas the standalone Fluent gets it normally!). Workbench starts, displays :
Local server exited or could not read local server port ANSYSLI_LOCAL_PORT
in a popup, and is then only usable for tasks that don't need a license, ie almost nothing...

for me the solution was to issue the following commands on the file server, from the directory containing ansys_inc :
sudo mv ansys_inc/shared_files/licensing/linx64/update/* ansys_inc/shared_files/licensing/linx64
sudo rmdir ansys_inc/shared_files/licensing/linx64/update

[mhsr]

Originally Posted by baserinia

Hi Everyone,

I have the same problem. Everything works fine but I cannot open the old .cfx file from version 11. I keep getting an error box like this in CFX-Pre:

-------------------------------------------------------------------------------------------------
p, li { white-space: pre-wrap; } ERROR
CCL validation failed with message:
Warning: Can't check parameter Message Type (/PHYSICS MESSAGES/PHYSICS MESSAGE:msg0000/Message Type): Can't find rule.
Warning: Can't check parameter Severity (/PHYSICS MESSAGES/PHYSICS MESSAGE:msg0000/Severity): Can't find rule.
Warning: Can't check parameter Message (/PHYSICS MESSAGES/PHYSICS MESSAGE:msg0000/Message): Can't find rule.
Warning: Can't check parameter Suppressed (/PHYSICS MESSAGES/PHYSICS MESSAGE:msg0000/Suppressed): Can't find rule.
-------------------------------------------------------------------------------------------------

Does anybody else has the same problem?

I have the same problem on the first of the following three different combinations of CPU+debian release (combos B and C are doing fine):
A) Xeon E5540 + debian 6.0 - x86_64
B) Xeon E5540 + debian 5.0 - x86_64
C) Athlon II X2 240 + debian 6.0 - x86_64

Solver runs on combo A as master are failing too (but it does run fine as slave in a parallel setup). The solver stops with the following error:

+--------------------------------------------------------------------+
 |                                                                    |
 |                    CFX Command Language for Run                    |
 |                                                                    |
 +--------------------------------------------------------------------+

 SIMULATION CONTROL:
 END
 
ccl2flow: * command language error *

Message: getChildList: unable to find the requested path

Context: returned by cclApi call


 +--------------------------------------------------------------------+
 |                An error has occurred in cfx5solve:                 |
 |                                                                    |
 | /ansys_inc/v130/CFX/bin/linux-amd64/ccl2flow.exe exited with       |
 | return code 3.                                                     |
 +--------------------------------------------------------------------+

According to strace output of such a solver run, the problem is caused when data from <casename>.def is read+processed+written to <casename>.dir/def.
Compared to a strace output on combo B files def and ccl in <casename>.dir/ differ dramatically in size.

Looks like there is some problem with the libs in debian 6.0 in combination with the Intel CPU. But right now, I have no idea where to start looking for the deeper cause of this problem.

[fivos]

I would like to ask another question regarding Workbench itself now, running on ubuntu. A colleague of mine will need it for his work, so he requested me to check if Ansys 13 workbench works on our PC.

The PC has Ubuntu 8.10 x64. Licensing is ok, through FlexLM, CFX and fluent modules operate normally, also I deleted the .ansys and .config/ANSYS folders from the Home directory, to avoid the QPixmap error.

Now the problem is that after workbench loads, it gives the following error :

Could not connect to local server :
24800@localhost

and there is no available analysis, even if we have licenses available (something like the problem that appears here but on windows 7 : http://www.cfd-online.com/Forums/ans...h-problem.html).

What this could be ? Does it have anything to Linux Standard Base ? (Unfortunately I cant download it through synaptic, or sudo apt-get since support has been dropped for Ubuntu 8.10, thus no files are found).

Also my hosts file is the following :

127.0.0.1 localhost
127.0.1.1 fivos-desktop


# The following lines are desirable for IPv6 capable hosts
::1 ip6-localhost ip6-loopback
fe00::0 ip6-localnet
ff00::0 ip6-mcastprefix
ff02::1 ip6-allnodes
ff02::2 ip6-allrouters
ff02::3 ip6-allhostsglibc-2.12-1
ldd --version

Is there anything with it ?

Any ideas are appreciated.

[ginggs]

Originally Posted by ffg

Application Error in CFD-Post
ERROR
CCL validation failed with message:
value 'Coordinate Frame' is not allowed for parameter /DEFAULT INSTANCE TRANSFORM: Default Transform/Rotation Axis Type

SEVERE
An attempt to recover from a previous error has failed. The application is likely to be in an unstable state

I have also seen this error on an Ubuntu 10.04 x86_64 installation, running CFD-Post from the terminal once stopped the error from appearing again, even when started from Workbench.

/ansys_inc/v130/CFD-Post/bin/cfdpost

Then things got very strange, on one PC with a Xeon processor and Nvidia Quadra graphics card we continued to get the above error when trying to view surfaces or create contour plots.
On another PC with Pentium D processor and Nvidia 6600 graphics card, everything worked fine.

We tried swapping graphics cards and the Xeon PC with the Nvidia 6600 card still produced the error.

We then tried moving the hard-drive containing the Ubuntu 10.04 x86_64 installation from the Pentium D PC to the Xeon PC, and the Xeon PC still produced the error.

[kruka]

1. First of all, install the prerequisites (may be not a full list):

sudo apt-get install xterm libstdc++6-4.4-dev \
libmotif-dev libxtst-dev libxt-dev libzip-dev libxmu-dev \
tcl8.5-dev tk8.5-dev

2. The previous authors suggested creating some custom scripts like uname, rpm and so on. This is not necessary. Instead, just install the Linux Standard Base package:

sudo apt-get install lsb

3. As someone already pointed out, for Nvidia cards starting from the driver version 260.x the creation of a symbolic link (ln -s /usr/lib/nvidia...) is not necessary as this is done by default. There is also no need to additionally define the LD_LIBRARY_PATH variable too.

4. Mount the ANSYS iso image, run the INSTALL script, install the license

5. If you installed ANSYS as root, you need to change ownership of the ANSYS configuration directories and files so that you have write access to them ("user" is the your user name):

sudo chown -R user:user ~/.ansys
sudo chown -R user:user ~/.config/Ansys

6. For convenience, add these lines to your ~./bashrc file:

alias launcher130='/ansys_inc/v130/ansys/bin/launcher130'
alias runwb2='/ansys_inc/v130/Framework/bin/Linux64/runwb2'

6.1 Then run:

source ~/.bashrc

7. Make sure the /etc/hosts file contains the right lines, as suggested by the authors above

That's it!

[zippyhybrid]

Has anyone tried doing this on a MacBook Pro? I've been dual-booting mine for the past year to run ANSYS natively in Windows 7 64-bit using Bootcamp, but I think I would prefer to switch to Linux, since I've had some frustrations with Windows.

This thread has been extremely helpful...thank you! As a trial run, I followed the advice of Yankee and kruka and got Workbench, DesignModeler, and Fluent working on ANSYS 13.0 installed in Ubuntu 10.10 running under Parallels 6. However, when I try to run the Mesh Editor, it crashes the OS and causes Ubuntu to restart. Also, before trying Ubuntu, I installed ANSYS 13.0 in CentOS 5 running under Parallels. The same components mentioned above worked fine right out of the box on CentOS. Mesh Editor would open fine, but when I actually tried to generate a mesh Workbench would crash, and Terminal would display something like "An internal error occurred" and no additional information.

Either way, I can't seem to get any specific information from the system as to why this crash occurs. My suspicion is that it may have something to do with the virtualization software but since neither Parallels nor ANSYS would officially support what I am trying to do, I can't really ask them about it. I am tempted to try and install Ubuntu on an external USB drive to see if the same thing happens when running it natively on the MacBook Pro, but I thought I would post the question here to see if anyone has any thoughts on the issue.

Thanks again...

[siden]

ansys works perfectly whith distros like opensuse (i tested it) ans fedora. so why not change to theses distros.they are very good

[titrisol]

Ihave installed Fedora 14 (fresh) and Ansys 13; but Design Modeler is not starting.
I see the red light saying Starting DesingModeler.... and then it sticks there for hours.

Fluent works fine; so does CFX and post-processing

I have tries serveral of the solutions posted here with no luck. Any ideas?

Also does anyone know how to make the Fluent fonts look better? the appear big

[Jimmy123]

With the instructions from Kruka I managed to install Ansys 13 on Ubuntu 10.10. Thank you very much Kruka!

Originally Posted by kruka

1. First of all, install the prerequisites (may be not a full list):

sudo apt-get install xterm libstdc++6-4.4-dev \
libmotif-dev libxtst-dev libxt-dev libzip-dev libxmu-dev \
tcl8.5-dev tk8.5-dev

2. The previous authors suggested creating some custom scripts like uname, rpm and so on. This is not necessary. Instead, just install the Linux Standard Base package:

sudo apt-get install lsb

3. As someone already pointed out, for Nvidia cards starting from the driver version 260.x the creation of a symbolic link (ln -s /usr/lib/nvidia...) is not necessary as this is done by default. There is also no need to additionally define the LD_LIBRARY_PATH variable too.

4. Mount the ANSYS iso image, run the INSTALL script, install the license

5. If you installed ANSYS as root, you need to change ownership of the ANSYS configuration directories and files so that you have write access to them ("user" is the your user name):

sudo chown -R user:user ~/.ansys
sudo chown -R user:user ~/.config/Ansys

6. For convenience, add these lines to your ~./bashrc file:

alias launcher130='/ansys_inc/v130/ansys/bin/launcher130'
alias runwb2='/ansys_inc/v130/Framework/bin/Linux64/runwb2'

6.1 Then run:

source ~/.bashrc

7. Make sure the /etc/hosts file contains the right lines, as suggested by the authors above

That's it!

However, I would like to share my experience with some additional problems that I got.
1: The workspace in design modeller was white. This was solved by typing

sudo ln -s /usr/lib/nvidia-current/libGL.so.1 /usr/lib/libGL.so.1

in the terminal.

2: Icemcfd did not start. I got an error message saying something like "unexpected "else", expected "fi". This was solved by typing

sudo apt-get install csh

in the terminal.

Edit: Oh no, I just found out that I have the CCL-problem when I am trying to start PRE.

[zippyhybrid]

I installed ANSYS 13.0 on Ubuntu 10.10 running natively on my Macbook Pro 5.3 (using the technique described above by kruka plus the additional step #1 posted by Jimmy123...Thank You!), and almost everything now seems to work...I think the errors I was having before with the Mesher were due to Parallels not properly providing a virtual graphics driver. However, I am also getting this error:

Originally Posted by ffg

------
Application Error in CFD-Post
ERROR
CCL validation failed with message:
value 'Coordinate Frame' is not allowed for parameter /DEFAULT INSTANCE TRANSFORM:Default Transform/Rotation Axis Type

SEVERE
An attempt to recover from a previous error has failed. The application is likely to be in an unstable state
------

I have isolated it to be a configuration problem with CFD-Post. 'Rotation Axis Type' is a parameter of the Default Transform that can either have the value 'Principle Axis' or 'Rotation Axis'. However, when CFD-Post starts up, it is given the value 'Coordinate Frame' and thus the error occurs. It can be changed manually in the Command Editor (right-click Default Transform in the sidebar and choose "edit in Command Editor") but this only seems to work if CFD-Post is opened as a standalone system, and performed before the results are loaded. The next time it is launched it defaults back to the erroneous value and the system is broken.

Is there some sort of startup setting that needs to be modified here so this doesn't happen?

[buck2202]

Originally Posted by zippyhybrid

I installed ANSYS 13.0 on Ubuntu 10.10 running natively on my Macbook Pro 5.3 (using the technique described above by kruka plus the additional step #1 posted by Jimmy123...Thank You!), and almost everything now seems to work...I think the errors I was having before with the Mesher were due to Parallels not properly providing a virtual graphics driver. However, I am also getting this error:



I have isolated it to be a configuration problem with CFD-Post. 'Rotation Axis Type' is a parameter of the Default Transform that can either have the value 'Principle Axis' or 'Rotation Axis'. However, when CFD-Post starts up, it is given the value 'Coordinate Frame' and thus the error occurs. It can be changed manually in the Command Editor (right-click Default Transform in the sidebar and choose "edit in Command Editor") but this only seems to work if CFD-Post is opened as a standalone system, and performed before the results are loaded. The next time it is launched it defaults back to the erroneous value and the system is broken.

Is there some sort of startup setting that needs to be modified here so this doesn't happen?

I verified on a Windows machine that 'Coordinate Frame' is the setting for /DEFAULT INSTANCE TRANSFORM: Default Transform/Rotation Axis Type when you open up a "results" node from Workbench. So, I did some searching through the files in the CFD-Post folder and made a change that (seems to) fix the issue.

1. Make a backup of the file that we're about to change..

$ sudo cp /ansys_inc/v130/CFD-Post/etc/CFXPostRules.ccl /ansys_inc/v130/CFD-Post/etc/CFXPostRules.ccl.orig

2. Edit CFXPostRules.ccl (as root... "sudo vi file", "gksudo gedit file" etc.) and change line 2626 FROM

   Allowed String List = Principal Axis, Rotation Axis

TO

   Allowed String List = Principal Axis, Rotation Axis, Coordinate Frame

For whatever reason, the CFXPostRules.ccl file from a Windows install doesn't allow "Coordinate Frame" either, but the application doesn't complain about it. I haven't tested this extensively, but I can at least launch the program and view results now. I'm running ANSYS 13 on Ubuntu 10.10 amd64.

[mahill]

I'm having an issue I believe someone else also addressed. DM and Meshing works well, but CFX-pre tries to start, but just the window hangs and you can't close the window.

[fabio moretti]

Bad news.
The same CCL validation failure messages experienced by many users on various distros appeared to me on OpenSUSE 11.4 too.

[mahill]

Originally Posted by fabio moretti

Bad news.
The same CCL validation failure messages experienced by many users on various distros appeared to me on OpenSUSE 11.4 too.

I'm surprised there is such difficulty with OpenSUSE 11.4 considering SUSE 11 is supposed to be an officially supported distro. I'm curious if it's a Motif vs. Open Motif difference. I believe that's one difference between SUSE and OpenSUSE.

[ginggs]

Originally Posted by ginggs

Then things got very strange, on one PC with a Xeon processor and Nvidia Quadra graphics card we continued to get the above error when trying to view surfaces or create contour plots.
On another PC with Pentium D processor and Nvidia 6600 graphics card, everything worked fine.

We tried swapping graphics cards and the Xeon PC with the Nvidia 6600 card still produced the error.

We then tried moving the hard-drive containing the Ubuntu 10.04 x86_64 installation from the Pentium D PC to the Xeon PC, and the Xeon PC still produced the error.

I assumed that it was just this one Xeon PC that had the problem, but since then as we have been upgrading to Ansys 13.0 a pattern emerged; only the PCs with quad core Intel CPUs produced this error, quad core AMDs and single core Intels had no problem.

I can confirm that buck2202's procedure below fixed the issue on our quad core Intels:

Originally Posted by buck2202

1. Make a backup of the file that we're about to change..

$ sudo cp /ansys_inc/v130/CFD-Post/etc/CFXPostRules.ccl /ansys_inc/v130/CFD-Post/etc/CFXPostRules.ccl.orig

2. Edit CFXPostRules.ccl (as root... "sudo vi file", "gksudo gedit file" etc.) and change line 2626 FROM

   Allowed String List = Principal Axis, Rotation Axis

TO

   Allowed String List = Principal Axis, Rotation Axis, Coordinate Frame

[fabio moretti]

I managed to have CFX 13.0 almost properly working on OpenSUSE 11.2, after unsuccessful attempts on Fedora (11, ..., 14), on Ubuntu 10.04 and 10.10, and on Opensuse 11.3 and 11.4.
So, no "CCL validation" error message any longer.

This is a good result for me, because on this distribution I can easily install also recent releases of some non commercial software.

"Almost", because I still have to fix a pvm-related problem, due to which I can't run distributed parallel calculations with the opensuse local machine and a fedora remote host (running the same code version).
Any hint on how to address this last problem would be much appreciated... Thanks.

Fabio

[mahill]

Originally Posted by fabio moretti

I managed to have CFX 13.0 almost properly working on OpenSUSE 11.2, after unsuccessful attempts on Fedora (11, ..., 14), on Ubuntu 10.04 and 10.10, and on Opensuse 11.3 and 11.4.
So, no "CCL validation" error message any longer.

This is a good result for me, because on this distribution I can easily install also recent releases of some non commercial software.

"Almost", because I still have to fix a pvm-related problem, due to which I can't run distributed parallel calculations with the opensuse local machine and a fedora remote host (running the same code version).
Any hint on how to address this last problem would be much appreciated... Thanks.

Fabio

Thanks for the info. That's good to know that OpenSUSE 11.2 works.

[charlotte]

Originally Posted by mahill

Thanks for the info. That's good to know that OpenSUSE 11.2 works.

I have CFX13.0 running on OpenSUSE 11.3. Most of the issues I got were related with my graphic card. Once I compiled the driver, and made sure it was installed properly (fglrxinfo), ANSYS was running well

[joeson789]

Hi,

I am using archlinux 64bit and I have successfully installed ANSYS 12.1 using the help in this thread, I can start workbench using the runwb2 script fine and the designmodeler seems to be working although it is very slow to start.

My main problem right now is the "Unable to create object WBScriptShell 121" error when I start the mesh editor, is there a solution or workaround to this problem?

Error: Unable to create object WBScript.WBWshShell.121. Please contact technical support.

Line: 3732
Char: 5
Error: Automation server can't create object
Code 800a01ad
Source: Microsoft JScript runtime error
Script: return AnsObjectFactory.CreateDispatchObject(progID);

I also get these lines in the terminal

igamma = 45455
igamma = 45455
igamma = 45455
igamma = 45455
igamma = 45455
igamma = 45455
igamma = 45455
igamma = 45455

Entered DesignModelerHelper::SetSystemName
Entered DesignModelerHelper::OpenAGDB
DM::OpenAGDB - Now register Imported files

************************************************** ****************************
*** A fatal error has occurred within AnsysWBU.exe and it must be closed...***
************************************************** ****************************

And finally I get this in WorkBench after the crash

Error! Unable to attach to geometry file /home/...../FFF.agdb. Tcp transport error.
Error! Plugin Error: Geometry Interface not found

Thank you very much.

PS: What is the difference between starting workbench using the workbench script in "ansys_inc/v121/aiso" and the runwb2 script in "ansys_inc/v121/Framework/bin/Linux64" ? It seems that when I run the workbench script, only the designmodeler is starting and I get the error "Workbench Error: error connecting to Albion". While when I start the script runwb2 I can start the designmodeler normally but I get the error aforementioned when I start the meshing editor.

[fabio moretti]

Originally Posted by charlotte

I have CFX13.0 running on OpenSUSE 11.3. Most of the issues I got were related with my graphic card. Once I compiled the driver, and made sure it was installed properly (fglrxinfo), ANSYS was running well

Do you mean that you didn't get any CCL validation error message when launching CFX-Pre? That's good to know. Thanks