OpendTect Administrators Manual

OpendTect Administrator's
Manual - 6.6
Copyright © 2002-2022 by dGB Beheer B.V.
All rights reserved. No part of this publication may be reproduced and/or published
by print, photo print, microfilm or any other means without the written consent of
dGB Beheer B.V.
Under the terms and conditions of any of the following three license types, license
holders are permitted to make hard copies for internal use:
l GNU GPL
l OpendTect Pro License
l Academic License
-1-
Table of Contents
OpendTect Administrator's Manual - 6.6 1
Table of Contents 2
System Requirements 6
Windows 6
Minimum 6
Recommended 6
For Machine Learning 7
Linux 8
Minimum 8
Recommended 8
Mac OS 12
Minimum 12
Recommended 12
Installation 14
OpendTect 14
Installing OpendTect on Windows 14
Installation via Installation Manager 14
Installer logging 21
Installing OpendTect on Linux 22
-2-
Required system libraries 22
lsb 23
xcb 24
Installer logging 30
Installing OpendTect on RHEL 6.x 31
Installing OpendTect on Mac OS X 33
Security setting 33
Licenses 41
Licensing and Host IDs 41
Installing demo/node-locked licenses 44
Installing floating licenses 48
Setting the DGBLD_LICENSE_FILE environment variable 48
Using the FlexNet lmpath tool 49
Clear license installation 50
Prepare offline installation packages 52
Install OpendTect via offline installation package 54
Setup Python Settings 56
Setup a Custom Python environment 56
Setup Distributed Computing 58
Tips & tricks 61
-3-
Setup the preferred GPU with NVIDIA Control Panel 63
Setup the preferred GPU with Windows settings 68
Setup NVIDIA GPU Performance Counters 76
Environment variables 78
Firewall Rules and Open Ports 80
Firewall rules For OpendTect 6.6 (Windows) 80
Open Ports 80
Python environments 82
Information about the OpendTect Python environments 82
Installing Python environments with the OpendTect installer 85
Installing Python environments from YAML files 86
Download the OpendTect Python YAML files 86
Install the environments 86
Cupti library 88
Scipy missing DLL dependencies bug 88
License server 90
Installing a license server on Linux 90
Starting a license server 91
Error messages 91
Client installation 93
Installing a license server on Windows 94
Starting a license server 98
-4-
Client installation 99
Petrel* Plugin 100
Installing the PIP file 100
Choosing the preferred Communication Port (TCP/IP) 100
PetrelDirect user documentation 101
Installation of PetrelDirect plugin in OpendTect and Petrel* 101
OpendTect 102
Petrel* 102
Troubleshooting 106
Log files 107
Settings files in .od folder 108
Debugging OpendTect 109
Appendix A - FlexNet Licenses Explained 110
-5-
System Requirements
Windows
Minimum
l Version: 10/11, older Windows versions may work, but are no longer tested.
l CPU: Intel/AMD, 64 bits
l GPU: Basic Intel graphics cards or Nvidia (e.g. recent GeForce/Quadro/NVS series)

graphics cards; AMD graphics cards may work.
l Memory: DDR4 memory, 16 GB of RAM, OpendTect itself needs at least 2 GB RAM.

Therefore, 16 GB will almost certainly be the absolute minimum.
l Storage: Hard Disk
Recommended
l Version: 10/11
l CPU: Intel/AMD processor with 64 bit support, 3+ GHz multi-core.

Note that OpendTect uses all processors if necessary. The more cores and speed, the
better. OpendTect will automatically use multiple threads in many situations. This
depends on the type of attribute, display, etc. We put a lot of effort to get time-con-
suming tasks multi-threaded.
l GPU: Nvidia (e.g. recent main-stream up to high-end GeForce series) graphics cards.
Quadro or NVS series cards could give the bit extra you want. In doubt, buy the best
GeForce card you can find. When buying a laptop make sure that it has a Nvidia chip-
set.
l Memory: DDR4 or DDR5 memory, on the safe side don't go for less than 32 GB.
Buy as much memory that you can afford and fits in the system. The big clients for
example use nothing less than 512 GB.
l Storage: SSD is best, other good options are Hard Disk and Network Drive.
This is usually under-valued, but it's often the crucial performance component. SSD
-6-
disks will give a tremendous boost in performance; essentially, data on SSD disks
loads almost as fast as pre-loaded, in-memory data. Performance could be miserable
if data needed to stream through (relatively) slow disks and/or networks.
For Machine Learning

l Version: 10/11
l CPU: Intel, 64 bits for when using Python environment Intel™ Math Kernel - MKL for
Machine Learning using CPU only. AMD, 64 bits should be fine when using the Python
Environment with CUDA 10.1 for Machine Learning on the GPU.
Ideally you want the system to be expendable to 4 GPUs. The CPU will need to sup-
port all GPUs. Important to look for is how many PCIe lanes the CPU supports and how
many PCIe lanes are needed for the system's number of GPUs and M.2 NVMe SSDs.
We recommend to get a CPU with at least 8 cores, 16 threads and 40 PCIe lanes.
l GPU: Nvidia, GeForce or Quadro series.

The GPU needs to be fast enough and able to fit the model and data batch in memory.
When in doubt choose the one with more memory. Other things to look for is the num-
ber of CUDA cores, tensor cores and GB memory bandwidth per second. We recom-
mend the following cards:
l Turing architecture cards (CUDA 10 and later)
l Nvidia GeForce RTX 2080 Ti with 11 GB DDR6 memory and 4352 CUDA Cores
l Nvidia Quadro RTX 6000 with 24 GB DDR6 memory and 4608 CUDA Cores
l Ampere architecture cards (CUDA 11.1 and later)
l Nvidia GeForce RTX 3090 with 24 GB DDR6 memory and 10496 CUDA Cores
l Nvidia A40 with 48 GB DDR6 memory and 10752 CUDA Cores
Buy as much memory that you can afford and fits in the system.
l Storage: The best choice is M.2 NVMe SSD that is big enough for the data.
The advantage of M.2 NMVe SSD is that it is plugged into the motherboard and is
super fast. Other options are SATA SSD, Hard Disk and Network Drive. Performance
could be miserable if data needed to stream through (relatively) slow disks and/or net-
works.
Please note that:
l For best performance OpenGL drivers should be up-to-date. For Machine Learning on

GPU we provide a Python package with CUDA 11.3. Please see this table on the
-7-
Nvidia CUDA Toolkit documentation page for the minimum compatible driver version.
l The CUDA 10 Python environment is now obsolete and will no longer receive security
updates. Users are encouraged to replace it with CUDA 11. Alternatively, you may
decide for the CPU-only Python environment.
l 4K/8K screens are not fully supported yet. This depends on the scaling factor. We are
working on a fix.
Please see the FAQ Visualization for a possible workaround.
l Windows needs to be updated with the latest updates from Microsoft.
Linux
Minimum
l Modern Linux distro.
l We have tested:
l RedHat/CentOS 7.2 and higher
l Ubuntu 16.04 and higher
l OpenSUSE Leap 15.1 and higher
l Other distros will probably work, possibly with as small tweak
l CPU: Intel/AMD processor with 64 bit support
l GPU: Basic Intel Graphics cards or Nvidia (e.g. recent GeForce/Quadro/NVS series)
graphics cards; AMD graphics cards may work.
l Memory: DDR4 memory, 16 GB of RAM, OpendTect itself needs at least 2 GB RAM.

Therefore, 16 GB will almost certainly be the absolute minimum.
Recommended
l Modern Linux distro. We have tested:
l CPU: Intel/AMD processor with 64 bit support, 3+ GHz multi-core.

Note that OpendTect uses all processors if necessary. The more cores and speed, the
-8-
better. OpendTect will automatically use multiple threads in many situations. This
depends on the type of attribute, display, etc. We put a lot of effort to get time-con-
suming tasks multi-threaded.
l GPU: Nvidia (e.g. recent main-stream up to high-end GeForce series) graphics cards.
Quadro or NVS series cards could give the bit extra you want. In doubt, buy the best
GeForce card you can find. When buying a laptop make sure that it has a Nvidia chip-
set.
Buy as much memory that you can afford and fits in the system. The big clients for
example use nothing less than 512 GB.
l Storage: SSD is best, other good options are Hard Disk and Network Drive.
This is usually under-valued, but it's often the crucial performance component. SSD
disks will give a tremendous boost in performance; essentially, data on SSD disks
loads almost as fast as pre-loaded, in-memory data. Performance could be miserable
if data needed to stream through (relatively) slow disks and/or networks.

l Modern Linux distro. We have tested:
l CPU: Intel, 64 bits for when using Python environment Intel™ Math Kernel - MKL for
Machine Learning using CPU only. AMD, 64 bits should be fine when using the Python
Environment with CUDA 10.1 for Machine Learning on the GPU.
Ideally you want the system to be expendable to 4 GPUs. The CPU will need to sup-
port all GPUs. Important to look for is how many PCIe lanes the CPU supports and how
many PCIe lanes are needed for the system's number of GPUs and M.2 NVMe SSDs.
We recommend to get a CPU with at least 8 cores, 16 threads and 40 PCIe lanes.
l GPU: Nvidia, GeForce or Quadro series.

The GPU needs to be fast enough and able to fit the model and data batch in memory.
When in doubt choose the one with more memory. Other things to look for is the num-
ber of CUDA cores, tensor cores and GB memory bandwidth per second. We recom-
mend the following cards:
l Turing architecture cards (CUDA 10 and later)
-9-
l Ampere architecture cards (CUDA 11.1 and later)
l Nvidia GeForce RTX 3090 with 24 GB DDR6 memory and 10496 CUDA Cores
l Nvidia A40 with 48 GB DDR6 memory and 10752 CUDA Cores
l Storage: The best choice is M.2 NVMe SSD that is big enough for the data.
The advantage of M.2 NMVe SSD is that it is plugged into the motherboard and is
super fast. Other options are SATA SSD, Hard Disk and Network Drive. Performance
could be miserable if data needed to stream through (relatively) slow disks and/or net-
works.
Please note that:
l OpendTect may work when using the Nouveau driver, however for best performance
the Nvidia driver should be installed. The nouveau driver does not support CUDA.
l For best performance OpenGL drivers should be up-to-date. For Machine Learning on

GPU we provide a Python package with CUDA 11.3. Please see this table on the
Nvidia CUDA Toolkit documentation page for the minimum compatible driver version.
l The CUDA 10 Python environment is now obsolete and will no longer receive security
updates. Users are encouraged to replace it with CUDA 11. Alternatively, you may
decide for the CPU-only Python environment.
l Low-level GPUs keep showing poor performances through the generations. Shading
functionality requires special GPU features, present in the main-stream and high-end
GeForce, Quadro and NVS cards. Nevertheless, under Linux, only Nvidia provides
drivers capable of running the shading feature. If you can't see any colors on graphic
elements, try disabling shading (Utilities > Settings > Look and Feel).
l 4K/8K screens are not fully supported yet. This depends on the scaling factor. We are
working on a fix.
Please see the FAQ Visualization for a possible workaround.
l Linux distros should be LSB compliant. You can check this using the command lsb_
release. This is particularly stringent for commercial plugins using the FlexNet sys-
tem. There is documentation on installing license files for commercial plugins, and
there is a page with background information.
- 10 -
l Linux 64 bits releases require the libstdc++ library to be present on the system. In the
table you can see the minimum libstdc++ library version that is needed:
OpendTect version libstdc++ library needed

6.4 6.0.19 or newer
6.6 6.0.21 or newer
l Linux distros will need to have the XCB libraries installed. For check and installation
instructions please see Installing OpendTect on Linux.
l OpendTect is known to work under Red Hat, CentOS, Ubuntu, OpenSUSE and other
distributions, as well as earlier versions of the main distributions, too. Fedora usage is
not recommended - although it may work it's the only distro that regularly fails to work
in combination with OpendTect. This is probably because the graphics vendors do not
support it well in terms of drivers.
- 11 -
Mac OS
Minimum
l Version: Mac OS X 10.14 (Mojave) and 10.15 (Catalina)
l CPU: Mac/Intel processor with 64-bit support
l GPU: Basic Intel or AMD graphics card, f.i. the Intel HD Graphics 4000.
l Memory: 16 GB of RAM
Recommended
l Version: Mac OS X 10.14 (Mojave) and 10.15 (Catalina)
l CPU: Mac/Intel processor with 64-bit support
l GPU: Intel or AMD graphics card
l Memory: Don't go for less than 32 GB RAM.

l Storage: SSD is best. Other good options are Hard Disk and Network Drive.

l This is still to be determined.
Please note:
l Machine Learning on GPU is not supported due to the absence of Nvidia cards for
recent Mac OS X systems.
l Mac/PowerPC support is NOT available.
l Mac/ARM support is NOT yet available.
l Mac/Intel emulation in Rosetta 2 is not supported.
- 12 -
l A 3-button mouse is highly recommended.
- 13 -
Installation
OpendTect
Installing OpendTect on Windows

OpendTect Pro and OpendTect Pro + Plugins (the commercial products) or
OpendTect (the free version) together with Python packages, offline OpendTect
and dGB Plugins documentation and Developer Tools can be installed via the
OpendTect Installation Manager (OpendTect_Installer_win64.exe) or via an offline
package.
Installation via Installation Manager

1. The Installation Manager is available for download via the appropriate platform link on
the download page of the dGB website.
- 14 -
Figure: OpendTect download page
2. The Installation Manager is a wizard to install/update the existing OpendTect (Latest /

Previous) releases. The release type field is used to select the release that is needed
to be installed/updated. The installer gives you the choices as seen below:
- 15 -
Figure: OpendTect Installation Manager
3. If it is a fresh installation then you may keep the proposed installation base directory or
choose your own. If it is an update then select its installation base directory. To install
commercial - and free products on the same system, the commercial and free products
must be placed in separate installation folders, e.g. C:\Program Files\OpendTectFree
and C:\Program Files\OpendTectPro. Now click Proceed.
Please note: installing in C:\Program Files\OpendTect will require administrative

rights. In case you want to prevent using the administrative rights then we suggest to
install into a folder on which a normal user has permissions, e.g. into C:\User-
s\username\OpendTect
4. Select the package type of OpendTect you want to install. Please note that this window
will only popup when installing a fresh OpendTect or OpendTect Pro installation.
Option OpendTect Pro will automatically select packages OpendTect Pro and
Miniconda3. OpendTect Pro is the extended version of OpendTect.
Option OpendTect Commercial will select all packages with exception that the lesser
preferred Python package will not be automatically selected.
Option OpendTect Free will automatically select OpendTect and Miniconda3.
OpendTect is a free open source seismic interpretation system and software devel-
opment platform.
Documentation and Development Tools packages are not automatically selected. It is

possible to add packages at a later time.
To read more about OpendTect licensing, please refer to our webpage of licensing
types.
Figure: OpendTect Installer Package Selection window
5. The OpendTect Installation Manager identifies the platform on which it is running. This
information is then anonymized prior to it being sent to OpendTect. We use this
anonymous data solely for the purpose of getting a picture of OpendTect usage and
- 16 -
thus improving our support capabilities.
6. The next window of the wizard is the OpendTect Package Manager. Multiple items can
be toggled on or off by checking the boxes. Optionally, the relevant package com-
bination could also be selected from the top list box.
Figure: OpendTect Package Manager
7. The installation manager will automatically recognize the previously installed version
at the selected path and will prompt it in the Installed version field. To read more about
a particular item in the list, select the item by clicking on it and read the description on
to the right panel. For example, Machine Learning:
- 17 -
Figure: Machine Learning information in right panel of OpendTect Installation Manager
8. Please note: In order to get an out of the box experience with the Machine Learning
plugin you should select the Python CUDA 11 (GPU) package for training on the Nvidia
GPU. It is also possible to select the Python MKL (CPU), with Intel™ Math Kernel
Library, environment for training on the CPU only. The base Miniconda3 Python envir-
onment will always be installed. More information about system requirements for
Machine Learning are available.
9. After selecting the packages for installation click on Proceed to start the installation.
You will first get a window that downloads the files from the download server. After
downloading the packages they will be unpacked to the installation path.
- 18 -
Figure: Example of downloading files and archive unpacker windows
10. If, for any reason, you should choose to abort the installation mid-download, you will
see the following window appear:
Figure: Stopped accessing download.opendtect.org
This gives you various options, including increasing the time-out from its default set-
ting, changing the download server or changing the Proxy settings. One thing that can
be tried is temporarily (partly) disable the Windows firewall to find out whether that
helps.
11. The installation will finish with giving an overview of the installed packages. Here is an
overview for the OpendTect Commercial installation:
- 19 -
Figure: Overview of installed packages
12. Only for OpendTect 6.6 and newer: After installation is finished the Manage Firewall
Program Rule window appears. This window gives the possibility to add rules for a
selection of OpendTect and Python executables to the Windows Firewall. For more
information please see the Firewall rules and ports chapter.
Please note that for adding rules to the firewall elevated rights are needed.
- 20 -
Figure: Manage Firewall Program Rule
13. To start OpendTect double click the desktop icon OpendTect 6.6.0 or the App
OpendTect 6.6.0 from the Start menu. Alternatively one can browse into the install-
ation directory and run od_main.exe from C:\Program
Files\OpendTect\6.6.0\bin\win64\Release.
Installer logging
The installer logs to a logfile. If something goes wrong during installation please look
into the following files:
l %TEMP%\username_od_instmgr_install_log.txt
l %TEMP%\username_od_instmgr_updcheck_log.txt
- 21 -
Installing OpendTect on Linux
OpendTect Installation Manager (OpendTect_Installer_lux64.sh) or via an offline
package.
Required system libraries

As written in the Linux System Requirements, OpendTect requires several system
libraries . To be able to successfully launch the OpendTect Installer and OpendTect
itself you have to make sure that these system libraries are present:
l lsb
l xcb
- 22 -
lsb
You can check this using the command lsb_release.
How to install lsb? This differs per distro:
l Debian/Ubuntu: sudo apt install lsb
l RHEL/CentOS/Rocky: yum install redhat-lsb
l OpenSUSE: zypper install lsb
- 23 -
xcb
How to check for xcb? This differs per distro:
l Debian/Ubuntu: sudo apt list --installed | grep xcb
l RHEL/CentOS/Rocky: yum list installed | grep xcb
l OpenSUSE: zypper search --installed-only xcb
How to install xcb? This differs per distro:
l Debian/Ubuntu: sudo apt install libx11- xcb1 libxcb- util1

libxcb1
l RHEL/CentOS/Rocky: yum install libxcb xcb-util xcb-util-image

xcb-util-keysyms xcb-util-renderutil xcb-util-wm
l OpenSUSE: zypper install libX11-xcb1 libxcb-util1 libxcb1

- 24 -
2. After downloading the installer open a terminal window.
3. In the terminal window change the directory to the directory where the installer has
been downloaded to.
4. Then run sh OpendTect_Installer_lux64.sh

Please note: to be able to run the installer csh or tcsh needs to be installed
5. This should unpack the OpendTect Installer into the /tmp directory and then run the
Installer.

- 25 -
must be placed in separate installation folders, e.g. /home/your_user-
name/OpendTectFree and /home/your_username/OpendTectPro. Now click Proceed.
Option OpendTect Commercial will select all packages with exception that the lesser
preferred Python package will not be automatically selected.
opment platform.

types.
- 26 -
Figure: OpendTect Installer Package Selection window
thus improving our support capabilities.
- 27 -
to the right panel. For example, Machine Learning:
Figure: Machine Learning information in right panel of OpendTect Installation Manager
12. Please note: In order to get an out of the box experience with the Machine Learning
plugin you should select the Python CUDA 11 (GPU) package for training on the Nvidia
GPU. It is also possible to select the Python MKL (CPU), with Intel™ Math Kernel
Library, environment for training on the CPU only. The base Miniconda3 Python envir-
onment will always be installed. More information about system requirements for
Machine Learning are available.
You will first get a window that downloads the files from the download server. After
- 28 -
Figure: Stopped accessing download.opendtect.org
This gives you various options, including increasing the time-out from its default set-
ting, changing the download server or changing the Proxy settings. One thing that can
be tried is temporarily (partly) disable the Linux firewall or SELinux to find out whether
that helps.
- 29 -
16. To start OpendTect change the terminal or Konsole directory into /home/your_user-
name/OpendTectPro/6.6.0 and then run ./start_dtect
Installer logging
The installer logs to a logfile. If something goes wrong during installation please look
into the following files:
l /tmp/your_username_od_instmgr_install_log.txt
l /tmp/your_username_od_instmgr_updcheck_log.txt
- 30 -
Installing OpendTect on RHEL 6.x
The OpendTect Installation Manager and 6.6 are not supported for RHEL 6.x. It is
however possible to install and run OpendTect 6.4 on RHEL 6.x and CentOS 6.x by
installing via an offline installation package. This offline installation package can be
created with the OpendTect Installation Manager. This can be done on a supported
Linux Distribution or Windows.
To install the offline package:
1. Transfer the od_offline_lux64 folder to the RHEL6.x or CentOS 6.x system
Then on the destination system in terminal:

2. cd od_offline_lux64
3. ./unix_install.csh
Please note: to be able to run the installer csh needs to be installed
4. The installer asks for the path of where to install OpendTect, e.g. /home/user-
/OpendTect
5. After installation is finished change the directory to the OpendTect installation directory
6. Run ./start_dtect to start OpendTect
On RHEL 6.x and CentOS 6.x there can be error messages that will prevent you
from starting OpendTect.
You might get:

/usr/lib64/libstdc++.so.6: version `GLIBCXX_ 3.4.14' not
found (required by /home/user-
/OpendTect/6.4.0/bin/lux64/Release/libosgQt5.so.145)
You might be able to solve this by installing a newer libstdc++ package. Here are
some instructions to get it installed. There are no guarantees these instructions will
work for your RHEL6.x or CentOS 6.x system. However, please note that it worked
on our CentOS 6.x system:
l wget https://rpmfind.net/linux/centos/7.8.2003/os/x86_64/Pack-
ages/libstdc++-4.8.5-39.el7.x86_64.rpm
l yum install libstdc++-4.8.5-39.el7.x86_64.rpm
- 31 -
When installing the newer libstdc++ package you will probably get the issue that you
need a newer libc.so.6. Here are some instructions to get it installed. There are no
guarantees these instructions will work for your RHEL6.x or CentOS 6.x system.
However, please note that it worked on our CentOS 6.x system:
l wget http://copr-be.cloud.fe-
doraproject.org/results/mosquito/myrepo-el6/epel-6-x86_
64/glibc-2.17-55.fc20/glibc-2.17-55.el6.x86_64.rpm
64/glibc-2.17-55.fc20/glibc-common-2.17-55.el6.x86_64.rpm
64/glibc-2.17-55.fc20/glibc-devel-2.17-55.el6.x86_64.rpm
64/glibc-2.17-55.fc20/glibc-headers-2.17-55.el6.x86_64.rpm
l yum install glibc-2.17-55.el6.x86_64.rpm glibc-common-2.17-

55.el6.x86_64.rpm glibc-devel-2.17-55.el6.x86_64.rpm glibc-
headers-2.17-55.el6.x86_64.rpm
- 32 -
Installing OpendTect on Mac OS X
OpendTect Installation Manager (OpendTect_Installer_mac.dmg) or via an offline
package.
Security setting
Please do the following before installation:
l Click on the 'Apple' button (top left), go to:
l System Preferences > Security & Privacy > General Tab
l In case the option is there then toggle on Allow apps to be installed from App Store and
identified developers, this may ask for admin credentials.
l Keep this window open when starting to install OpendTect. When you start the installer
for the first time you might get the error message: "OpendTect Installer" can't be
opened because the identity of the developer cannot be confirmed. or the error mes-
sage "OpendTect Installer" can't be opened because Apple cannot check it for mali-
cious software.
At System Preferences > Security & Privacy > General Tab there should be a
message like "OpendTect Installer" was blocked from opening because the identity of
the developer" cannot be confirmed. or "OpendTect Installer" was blocked from use
because it is not from an identified developer. Click the lock icon to make changes and
then click the Open Anyway button.
- 33 -
Figure: Set Security setting to App Store and identified developers and Open OpendTect Installer any-
way
l Once the installation is finished, and assuming you made a change to the security set-
ting then you may change the security back.
Please note that in order to install OpendTect into the Applications folder the user
that installs OpendTect should be allowed to administer this computer. You can
check whether this option is toggled on via Users & Groups.
- 34 -
Figure: Find out whether user is allowed to write into Applications folder via Users & Groups

- 35 -
2. Run the .dmg file via Finder.
3. Then start the OpendTect Installer.
Figure: Start the OpendTect Installer by opening the .dmg file
4. In case you get a window with error message: "OpendTect Installer" can't be opened
because the identity of the developer cannot be confirmed. or "OpendTect Installer"
can't be opened because Apple cannot check it for malicious software then please fol-
low the Security settings section in the top of this chapter to allow the opening of the
OpendTect Installer.
- 36 -
Figure: Error messages explaining that the OpendTect Installer can't be opened

must be placed in separate installation folders, e.g. /Applications/OpendTectFree and
/Applications/OpendTectPro. Now click Proceed.
Option OpendTect Commercial will select all packages.
opment platform.
- 37 -
types.
Figure: OpendTect Installer Package Selection Window
thus improving our support capabilities.|
- 38 -
to the right panel.
12. You may get a window with error message You do not have write permission in /Ap-
plications/OpendTect. This means that the current user has no administrative rights to
install into the Applications folder. The solution is to either give the current user admin-
istrative rights or to install into a folder where the user is allowed to write files, like e.g.
the users Documents/OpendTect folder.
Figure: No write permissions in /Applications/OpendTect
13. You will first get a window that downloads the files from the download server. After
This gives you various options, including increasing the time-out from its default
- 39 -
setting, changing the download server or changing the Proxy settings.
16. To start OpendTect open it with Finder: Applications > OpendTectPro > OpendTect
6.6.0
- 40 -
Licenses
Licensing and Host IDs

OpendTect uses FlexNet to manage its licensing. There are two main types of
license:
l Node-locked: The license file is tied directly to a specific client machine (or selection
of machines) through its ‘HostID’. This option is popular for laptops and for single-user
desktop setups. Installation is very simple.
l Floating License: The license file is generated for, and tied to a dedicated server via
its ‘HostID’. The license manager on the server then issues licenses to client
machines. This option is more suited for multiple users. Installation can be more
involved.
In order to generate a license, we need the HostID (commonly known as MAC

address) of the machine or server, depending on license type:
l If server-based (floating) license

l Server (Host) Name
l Server (Host) ID
l If node-locked license
l Host ID
To discover the HostID(s):
- 41 -
This will give a result similar to the following window:
Use the clipboard icon to copy the information displayed in the window
into text format for pasting into an email or document.
Note: On some Windows 10 and 11 systems there is an option Random hardware

addresses in Windows Settings > Network & Internet > WiFi . Please make sure that
the Use random hardware addresses option is toggled off when using the Show
HostID utility and supplying us with the information needed for generating the
license. Also this option should be toggled off when using the OpendTect license
that you received back. When a random HostID is supplied to us we can not guar-
antee that the generated OpendTect license will keep working. Therefore we ask
you to supply us with a HostID that will not change.
- 42 -
Figure: Random hardware addresses option in Windows Settings > Network & Internet > WiFi
For additional information (including alternate methods of accessing the HostID),

please refer to the Appendix FlexNet Licenses Explained.
- 43 -
Installing demo/node-locked licenses
Under Utilities > Installation > Licenses you will see two sets of options, differing per
platform:
Figure: License options on Windows
Figure: License options on Linux
- 44 -
Windows:
Clicking the option Install/Edit licenses... will bring up the following tool:
Linux:
Clicking the option Install/Edit licenses... will bring up the same tool but note that the
License Manager Tools button is not present (this is a Windows-only option):
For both Windows & Linux:
As shown above, select File from the drop-down and Select to browse to the node-
locked license to be installed.
You may then Examine the license:
- 45 -
If everything seems in order, proceed to Install the license. On pressing Install, you
will see a warning that a restart will be necessary:
Pressing Continue will show this message:
- 46 -
Clicking OK will prompt OpendTect to restart to complete the installation of the
node-locked license.
For information about floating or server- based licenses, please refer to the
Appendix FlexNet Licenses Explained or look in the chapter installing floating
licenses for instructions on how to get them installed on the client.
For more general information about OpendTect licensing options, please see the
licenses page
A more complete explanation of OpendTect license installation can be found in the

License Installation Webinar, available on OpendTect's YouTube Channel.
- 47 -
Installing floating licenses
A floating license file is to be used by the dGB FlexNet license server. This license
server can run on Linux, Mac or Windows. See here for instructions on how to install
the license server on Linux and how to install the license server on Windows.
Once the license server is running we can then instruct the client on how to borrow
the license. There are two ways for installing the floating license on a client:
1. By setting the DGBLD_LICENSE_FILE environment variable
2. By using the FlexNet lmpath tool via console/command prompt or on Windows

through the License Manager Tools (lmtools.exe)
Setting the DGBLD_LICENSE_FILE environment variable

The DGBLD_LICENSE_FILE environment variable can be set to a license file or to
the server host.
On a Linux client the variable can be set with the export or setenv command.
On a Windows client the variable can be set via Control Panel

> System > Advanced system settings > Environment Variables. Please note that a
user environment variable is only available for the user who sets the variable. It
might be a better choice to set the variable as a system environment variable. It then
can be globally accessed by all users.
The possible values are:
l port@host
l @host
l path to license file
- port = the TCP/IP port the license server is using. If port is not specified in front of
@home a default TCP/IP port number in the range of 27000- 27009 is used.
- host = the hostname of the license server. This comes from the SERVER line in
the license file.
- path to license file = exactly that. Place the license file in a network location avail-
able to all machines running OpendTect or copy it to all of the machines running
OpendTect (Typically: C:\Licenses)
- 48 -
Using the FlexNet lmpath tool
The lmpath tool can be used to add to, override, or get the current license path set-
tings.
This and other FlexNet tools are packaged in a single executable lmutil or lmu-
til.exe.
OpendTect 6.6.0 ships these tools. They can be found in the following paths:
l For Linux: OpendTect/6.6.0/bin/lux64/lm.dgb
l For Windows: OpendTect\6.6.0\bin\win64\lm.dgb
To set it on Linux run:

./lmutil lmpath -override dgbld port@host or ./lmutil lmpath -
override dgbld licensefile
On Windows run:
lmutil.exe lmpath - override dgbld port@host or lmutil.exe
lmpath -override dgbld licensefile
Note: lmpath sets the file $HOME/.flexlmrc on Linux and the FlexNet Licensing
registry entry on Windows platforms.
To display the current license path setting, use on Linux the command:
./lmutil lmpath - status
Or on Windows:
lmutil.exe lmpath -status
For a status of the currently in use licenses, use on Linux the command:
./lmutil lmstat - a - c port@host
Or on Windows:
lmutil.exe lmstat -a -c port@host
- 49 -
Clear license installation
Under Utilities > Installation > Licenses you will see the option Clear license install-
ation:
Figure: Clear license installation option
This option (Windows only) will clear:
l Demo or node-locked licenses installed via any route, including the 'Install demo
license' option.
l Floating (or 'server') licenses that may have been installed (without stopping the
license server).
You will get this window explaining that all commercial licenses will be deacivated
and that the plugins will stop working. To remove click Remove or else Cancel.
- 50 -
Figure: Are you sure to remove the licenses?
Once cleared, you will be prompted to restart:
Figure: All installed licenses cleared
Users of Linux systems wishing to clear their license installation will need to do the
following:
l Locate the .flexlmrc file in your HOME directory (eg: $HOME/.flexlmrc)
l Check in the file for specific lines referring to the OpendTect vendors (dGB,
ARKCLS)
l If the file contains lines relevant to other software, then just delete the individual
lines. Otherwise, you may choose to delete the file.
This method also applies to both demo/node-locked and floating licenses and will
also not stop the server.
- 51 -
Prepare offline installation packages
Offline installation packages can be created with the OpendTect Installation Man-
ager. It is useful to do this if you want to install OpendTect on a system that has no
access to Internet. Please follow these steps to create such offline installation pack-
ages:
1. Download the OpendTect Installation Manager for Linux or Windows from the
OpendTect download page.
2. Run the installer
a. On Linux in the terminal run sh OpendTect_Installer_lux64.sh
b. On Windows simply double click on OpendTect_Installer_win64.exe
3. In the Installation Manager window select Prepare offline installation packages.
a. On Linux the default path of where the od_offline_lux64 or od_offline_win64

Download directory will be created will be /tmp/ . On Windows this is the user's
AppData\Local\Temp. It is possible to select another path.
b. At platform choose between Linux (64 bits) or Windows (64 bits).
4. Click the Proceed button to start creating the offline installation package.
Figure: Prepare offline installation packages
- 52 -
5. The rest of the setup is identical to the usual way of installing OpendTect. You
pick the packages that need to be included in the installation and click the Pro-
ceed button.
6. When the offline installation package has been created this will be shown in the
information window.
Figure: OpendTect offline-installation packages created successfully
7. The offline installation folder can now be transferred to the offline system by copy-
ing the folder via an USB stick or some other way onto the offline system.
- 53 -
Install OpendTect via offline installation package
On Linux:
1. Transfer the od_offline_lux64 folder to the offline system.
Then on the offline system in terminal:

2. cd od_offline_lux64
3. ./unix_install.csh
Please note: to be able to run the installer csh needs to be installed
4. The installer asks for the path of where to install OpendTect, enter it, e.g. /home/user-
/OpendTect
5. The questions about a previous installed version of OpendTect and the initial Project
data folder can be answered or ignored.
6. After a while the installation will be finished.
7. If the Python packages have been installed then in the terminal run the following to pre-
pare the Python environment(s) on the system:
cd OpendTect/Python
./setup.Python
8. OpendTect can be started by running the start_dtect script, which can be found in
the OpendTect/6.6.0 folder.
On Windows:
1. Transfer the od_offline_win64 folder to the offline system.
2. With the file explorer go into the od_offline_win64 folder.
3. Run OpendTect_setup.exe
- 54 -
4. Set the destination folder and press the Install button to start the installation.
Figure: Install OpendTect with the offline installation package on Windows
5. After a while the installation will be finished.
6. OpendTect can be started by running the od_main.exe executable, which can be

found in the OpendTect\6.6.0\bin\win64\Release folder.
- 55 -
Setup Python Settings
For information about the Python environments and how to install them please see
the following chapters:
l Information about the OpendTect Python environments
l Installing Python environments with the OpendTect installer
l Installing Python environments from YAML files
Setup a Custom Python environment

To use the environment that has been installed with the YAML file or if you have set
a custom Python environment in some other way, please do the following:
1. Start OpendTect 6.6
2. From the menu choose Utilities > Installation > Python Settings
3. Set Python environment to Custom
4. As Custom environment root choose the root directory of your own miniconda3
or anaconda3 installation, e.g. C:\ProgramData\Anaconda3.
5. With anaconda3 it is possible to install the environments into a different path than
the root directory. From OpendTect 6.6.3 and onwards OpendTect will allow this
by reading out the paths from %UserProfile%\.conda\environments.txt on Win-
dows or $HOME/.conda/environments.txt on Linux.
6. As Virtual environment choose odmlpython- cpu- mkl or odmlpython- cuda113.

The virtual environments are shown as full paths when they have been read from
the environments.txt file.
- 56 -
Figure: Set Python Settings window with a selected custom Python environment
7. If you want to use a Python IDE you can choose Jupyter-Lab, Jupyter-Notebook,
Spyder or Idle that come with the environment or choose your own IDE.
8. Press OK to save the Python settings.
- 57 -
Setup Distributed Computing
In order to utilize OpendTect's capability for Multi-Machine Processing (MMP), a
BatchHosts file must be created and used. This file contains the list of remote
machines ( host machines or nodes ) and some relevant details about these
machines and the path to the Survey Data Root. OpendTect will use this file to com-
municate to the remote hosts and launch processes remotely on them. Follow the
example format (shown below) to add the list of remote machines and their details in
the respective fields.
In order to minimize complications, the Setup Distributed Computing tool (formerly

known as Setup Batch Processing tool) can be used to create a tailor- made
BatchHosts file (via Utilities > Installation > Setup Distributed Computing...).
Figure: Launching Setup Distributed Computing tool
As default OpendTect will try to create a new or edit the existing BatchHosts file in
it's 6.6 .0/data directory. If this directory is not writable OpendTect will advise to
launch this process with administrator rights.
- 58 -
Figure: Selected Batch Host Directory is not writable
It is also possible to use a custom BatchHosts filepath by setting environment vari-

able DTECT_BATCH_HOSTS_FILEPATH .
Figure: Setting DTECT_BATCH_HOSTS_FILEPATH environment variable
Figure: Setup Distributed Computing window
BatchHosts file: This field is not editable in the User Interface.
IP address: IP address of the node machine(s). If the background of this field is in

red then there is a problem with the resolving of the hostname into the IP address.
- 59 -
Hostname: Hostname of the node machine(s). If the background of this field is in
red then there is a problem with the resolving of the hostname.
Display name: Free-text field. Text entered here appears in the Distributed Com-
puting window.
Platform: Select platform type, the options are: Linux (64 bits), Windows (32 bits),
Windows (64 bits) and Mac OS X.
Survey data root: Location of the survey (the path to the survey data root folder
from the host machine)
Figure: Advanced Settings window
Advanced Settings:
l Here you may change the first port value (in the case that it is blocked or in use). By
default this first TCP port is 37500. We advise to open up to 5 ports, e.g. 37500-37504.
l Linux users may decide to change the remote shell command from the default ssh to
rsh.
When setting to ssh it is required that the user who is running OpendTect is able to
login to the other nodes without a password. This can be done by setting up public key
authentication between the nodes. We will not go into detail of how to do this. In short
this is done by generating the SSH key on the machine you are using to start the jobs,
- 60 -
the public key then needs to be uploaded to the nodes and added to the user's .ssh/au-
thorized_keys file.
l The Nice level sets the priority on the host machines, 19 being nicest and 1 being least
nice).
l Finally, the Default Data Root can be set per platform.
Description of icons:
Add new host.
Remove selected host.
Move host up or down.
Test hosts. Will perform tests to ensure that the server and nodes can com-
municate to the necessary extent to perform the MMP. (ie: can the nodes find the
data root folder and read/write into it)
Tips & tricks

In case you are having issues with the Distributed Computing between the server
that launches the scheduler and the nodes please check the following:
l Can you ssh into the node(s) without having to type a password? e.g. run: ssh
nodehostname ls -la
l Can you the access the data project on the node(s)? e.g. run: ssh nodehostname
ls -la /pathto/dataproject
l Can you access OpendTect on the node(s)? e.g. run: ssh nodehostname ls -la

/pathto/opendtect
l When you run the scheduler is it listening on the port you set in the BatchHosts
file? e.g. run: netstat -antpu | grep 37500
- 61 -
l Is their a time difference on the server that is starting the scheduler and node(s)?
It is a good idea to sync them via a ntp server.
l Is the hostname resolution correct? Do the nodes get the correct IP address to
which they need to connect?
For more information on this topic, please refer to OpendTect's Youtube Channel
where you may find the webinar: Multi-Machine Processing Setup.
- 62 -
Setup the preferred GPU with NVIDIA Control
Panel
Laptops with an Intel CPU usually come with two integrated graphic cards. The
default one the laptop starts up with would be one such as Intel HD graphics.
OpendTect however will have the best performance when using the Nvidia GPU.
Therefore that one should be used instead.
When OpendTect starts it will scan your graphics card. If it finds an Intel card there
will be a warning message: 'Intel card found. If your computer has multiple graphics
cards, consider switching from the integrated graphics.'
Figure: Warning message about the Intel card
On Windows 8.1 and 10 up to feature release 2004 the Nvidia GPU can be made
the preferred GPU with the NVIDIA Control Panel. We assume that you have
installed the latest Nvidia driver that is available for your GPU. If not then please go
to the Nvidia website and download and install the latest driver for your type of GPU.
In case you are running Windows 10 feature release 20H2 or newer the preferred
GPU has to be set with the Windows Settings.
Please do the following:
- 63 -
1. Search for NVIDIA Control Panel, then start it
Figure: Starting NVIDIA Control Panel on Windows 10
2. In the toolbar on the left choose Manage 3D settings
Figure: NVIDIA Control Panel: 3D Settings > Manage 3D settings
- 64 -
3. At Tab Program Settings click the Add button
Figure: Add a program for customization of the 3D settings
4. In the new window it should show you OpendTect as a recently used program. Select it
and click on Add Selected Program.
Figure: List of recently used programs in NVIDIA Control Panel, add OpendTect
- 65 -
5. Select the preferred graphics processor: High-performance NVIDIA processor.
Figure: Customize OpendTect to use the NVIDIA GPU
6. Click Apply to confirm the OpendTect program customization.
7. The next time you start OpendTect it should no longer show the Intel card found mes-
sage popup. To check the graphics card used in OpendTect you might want to check
Utilities > Installation > Graphics Information. It should now show the Nvidia card
information and driver version.
- 66 -
Figure: Check which graphics card OpendTect uses
- 67 -
Setup the preferred GPU with Windows settings
Laptops with an Intel CPU usually come with two integrated graphic cards. The
default one the laptop starts up with would be one such as Intel HD graphics.
OpendTect however will have the best performance when using the Nvidia GPU.
Therefore that one should be used instead.
When OpendTect starts it will scan your graphics card. If it finds an Intel card there
will be a warning message: 'Intel card found. If your computer has multiple graphics
cards, consider switching from the integrated graphics.'
Figure: Warning message about the Intel card
From Windows 10 feature release 20H2 and onwards and Windows 11 the Nvidia
GPU can be made the preferred GPU with the Windows settings. We assume that
you have installed the latest Nvidia driver that is available for your GPU. If not then
please go to the Nvidia website and download and install the latest driver for your
type of GPU.
In case you are running Windows 8.1 or 10 feature release 2004 or older the pre-
ferred GPU has to be set with the NVIDIA Control Panel.
Please do the following on Windows 10:
- 68 -
1. Click the Start icon and then go to Settings > System
Figure: Navigating to the Windows 10 System settings
2. Windows settings should open the display settings. Browse down to Graphics settings
and click it.
Figure: Display > Graphics settings
3. Now choose an app to set the preference. OpendTect is a desktop app, so make sure
that the drop down bar has "Desktop app" selected. Now click the "Browse" button and
navigate to the OpendTect installation directory,
e.g.: C:\Program Files\6.6.0\bin\win64\Release and add od_main.exe
- 69 -
Figure: Choosing od_main.exe to set the GPU preferences
4. Now click the "Options" button and choose the High performance option and clicking
the "Save" button for setting the Nvidia GPU as default for OpendTect.
Figure: Setting the Nvidia card as high performance graphics preference for OpendTect
- 70 -
Please do the following on Windows 11:
1. Click the Start icon and then go to Settings > System > Display
- 71 -
Figure: Navigating to the Windows 11 System > Display settings
2. Windows settings should open the display settings. Browse down to Graphics settings
and click it.
- 72 -
Figure: Display > Graphics settings
3. Now choose an app to set the preference. OpendTect is a desktop app, so make sure
that the drop down bar has "Desktop app" selected. Now click the "Browse" button and
navigate to the OpendTect installation directory,
e.g.: C:\Program Files\6.6.0\bin\win64\Release and add od_main.exe
- 73 -
Figure: Choosing od_main.exe to set the GPU preferences
4. Now click the "Options" button and choose the High performance option and clicking
the "Save" button for setting the Nvidia GPU as default for OpendTect.
- 74 -
Figure: Setting the Nvidia card as high performance graphics preference for OpendTect
- 75 -
Setup NVIDIA GPU Performance Counters
When running jobs with the Machine Learning plugin you might get the following
error message: CUPTI_ERROR_INSUFFICIENT_PRIVILEGES
To fix this please follow the instructions from the NVIDIA site. It will be a good idea
to do this anyway to prevent of getting the above mentioned error message.
In short the way to set this on Windows is:
1. Open the NVIDIA Control Panel
2. From the menu select Desktop and ensure that Enable Developer Settings is checked
3. From the tray select Developer > Manage GPU Perfomance Counters
4. Select Allow access to the GPU performance counter to all users to enable unres-
tricted profiling
- 76 -
- 77 -
Environment variables
There are several environment variables that can be set. OpendTect will look for
these environment variables. They can be set to:
l let OpendTect use the dGB license from the FlexNet license server,
l get special features working,
l or to use custom paths for certain functionality.
Environment vari- Expected Values Description

able
DGBLD_LICENSE_ port@host Used on clients to use a dGB
FILE FlexNet server license.
f.i. - port = the TCP/IP port the license
27005@myserver server is using
- host = the hostname of the license
server
@host Used on clients to use a dGB
FlexNet server license.
f.i. @myserver - host = the hostname of the license
server
- port = not specified. A default TCP/IP
port number in the range of 27000-
27009 is used.
path to license file Used on clients to use a node-locked or
floating license.
- path to license file = exactly that. Place
the license file in a network location
available to all machines running
OpendTect or copy it to all of the
machines running OpendTect
(Typically: C:\Licenses)
DTECT_BATCH_ Path to a This sets the path to a tailor-made
HOSTS_FILEPATH BatchHosts file BatchHosts file. If it is not set then
OpendTect will use the default
BatchHosts file in the OpendTect install-
ation directory, e.g. C:\Program
Files\OpendTect\6.6.0\data\BatchHosts
.
- 78 -
DTECT_CLUSTER_ yes or no When set to yes OpendTect will have
PROC cluster processing enabled.
DTECT_PETREL_ Value of preferred This sets the preferred port for the
PORT port for the OpendTect Petrel* plugin access.
OpendTect
Petrel* plugin
access.
DTECT_SETTINGS Path to the This sets the path to the OpendTect set-
OpendTect settings tings home directory. Use this if you do
home directory not want to use the .od directory in the
user's home folder. Use this envir-
onment variable for Linux and Mac.
DTECT_ Path to the This sets the path to the OpendTect set-
WINSETTINGS OpendTect settings tings home directory. Use this if you do
home directory not want to use the .od directory in the
user's home folder. Use this envir-
onment variable for Windows.
LM_LICENSE_FILE port@host or Although it works it is preferred to set
@host or path to DGBLD_LICENSE_FILE instead.
the license file
OD_USE_ yes or no When set to yes OpendTect will support
VIRTUALMEM using virtual memory.
RSFROOT Path to the install- This sets the path to the Madagascar
ation directory of installation directory. If it is not set then
Madagascar and one would try to start Madagascar
from OpendTect an error message
would be displayed.
* Petrel is a mark of Schlumberger
- 79 -
Firewall Rules and Open Ports
Firewall rules For OpendTect 6.6 (Windows)
The installer will ask to add firewall rules to the Windows Firewall. Which .exe pro-
grams it adds to Windows Firewall depends on the packages that have been
installed. Here is the full list of .exe files that require a rule in the Inbound Rules in
the Windows Firewall. You may also do this manually. The default OpendTect base
directory is C:\Program Files\OpendTect. The base directory can be another dir-
ectory if that is chosen during installation. If that is the case then replace the default
base directory in below list with the actual base directory.
l C:\Program Files\OpendTect\6.6.0\bin\win64\Release\od_main.exe
l C:\Program Files\OpendTect\6.6.0\bin\win64\Release\od_SeisMMBatch.exe
l C:\Program Files\OpendTect\6.6.0\bin\win64\Release\od_remoteservice.exe
l C:\Program Files\OpendTect\Python\python.exe
l C:\Program Files\OpendTect\Python\envs\odmlpython-cpu-mkl\python.exe
l C:\Program Files\OpendTect\Python\envs\odmlpython-cuda113\python.exe
Open Ports
OpendTect will need open TCP ports on the localhost for Batch Processing and
Machine Learning. Here is a list of ports that OpendTect needs to open if you run a
certain job:
Ports Description
5050 Is used by executable od_remoteservice, which needs to be
started on the compute node before job is submitted to it, but
only when the job is started from Windows.
20050 - 20080 Starting from OpendTect 6.6 these ports are used for com-
munication between OpendTect and Python, starting at port
20050.
We advise to have a broad range to be open, e.g. until 20080.
37500 - 37504 Ports used for the Distributed Computing, starting at port
37500. We advise to have a port range size of 5 open.
That allows a user to start several Distributed Computing jobs
from the same client machine.
57375 Is used for the OpendTect Petrel* plugin access.
- 80 -
When using the FlexNet license server for the OpendTect license(s) it will need
open ports for the license server and vendor daemon(s). Here is a list of ports that
need to be open on the server that is running the FlexNet license server:
Ports Description
27000 - 27009 If no TCP port is specified in the license file the FlexNet license
or server will use an unused port in the range of 27000 - 27009.
specified port
random ephemeral If no TCP port is specified in the license file the dgbld and ark-
port clsld vendor daemons will listen on an ephemeral random port.
or
specified port
- 81 -
Python environments
Information about the OpendTect Python envir-
onments
Miniconda3 is a basic Python environment. It allows using the Presentation Maker
plugin without having to install Python yourself. Provides libraries, including:
l Python 3.8
l pandas
l python-pptx
Math Kernel Library (MKL) is a complete Python environment for Machine Learn-
ing using CPU only (with Intel™ Math Kernel Library - MKL).
Provides libraries, including:
l Python 3.8
l intel-tensorflow
l keras
l scikit-learn
l pandas
l bokeh
l numpy
l h5py
l joblib
l psutil
l xgboost
- 82 -
l python-pptx
l jupyterlab
l onnxruntime
l pytorch
l matplotlib
Python CUDA 11 is a complete Python environment for Machine Learning on the

GPU with CUDA 11.3.
Requires minimum driver version 466.XX, for specifics see this table on the Nvidia
CUDA Toolkit documentation page .
Will fallback automatically on the CPU if the hardware/driver requirements are not
met.
Provides libraries, including:
l Python 3.8
l tensorflow
l keras
l scikit-learn
l pandas
l bokeh
l numpy
l h5py
l joblib
l psutil
l xgboost
l python-pptx
- 83 -
l jupyterlab
l onnxruntime
l pytorch
l matplotlib
Please note: The Python CUDA 10 environment is now obsolete and will no longer
receive security updates. Also note that OpendTect does not support the Python
CUDA 10 environment from 6.6.6 and onwards.
- 84 -
Installing Python environments with the
OpendTect installer
When running the OpendTect installer there is the option to install the Python envir-
onments. The installer will auto select the Miniconda3 environment and also auto
select the Python CUDA 11 environment for training on the Nvidia GPU. For training
on CPU only the Math Kernel Library (MKL) Python environment should be selec-
ted. The Python CUDA 10 environment is now obsolete and will no longer receive
security updates. Users are encouraged to uninstall it, and replace it with either
CUDA 11 or CPU only Python environment. Also note that OpendTect does not sup-
port the Python CUDA 10 environment from 6.6.6 and onwards.
Figure: Install Python environments with the OpendTect installer
If you have your own miniconda3 or anaconda3 environment you might want to con-
sider to install the OpendTect Python environments from YAML files instead.
- 85 -
Installing Python environments from YAML files
The Python environments we provide can be installed into:
l the Python miniconda3 base directory that is installed with the OpendTect
installer,
l your own miniconda3 installation or into
l your own anaconda3 installation.

Download the OpendTect Python YAML files
Version 1.4.0 is the current version.
For Linux:
l odmlpython-cpu-mkl_lux64.yml
l odmlpython-cuda113_lux64.yml
For Windows:
l odmlpython-cpu-mkl_win64.yml
l odmlpython-cuda113_win64.yml
The YAML files contain a list of essential packages that are needed inside the
Python environment. You can choose to install either the CPU only (odmlpython-
cpu-mkl) or CUDA 11 one or both. In most cases the CUDA 11 one would be the
best option so that the Machine Learning jobs can be run on the Nvidia GPU.
Install the environments

On Linux:
1. Open a terminal/konsole.
2. Then run the following:
- 86 -
a. cd conda installation directory (e.g. cd /home/user/Miniconda3)
b. cd bin
c. bash ; source activate
d. conda env create --file /pathto/odmlpython-cpu-mkl_lux64.yml
e. conda env create --file /pathto/odmlpython-cuda113_lux64.yml
f. Conda will collect the packages it needs to install from the YAML file.
g. Then it will download and extract the packages, do the transaction and install
remaining packages with pip.
On Windows:
1. Open the Command Prompt. Run as administrator if the Python installation dir-
ectory is on a location for which the user requires administrative rights.
2. Then run the following:
a. cd conda installation directory (e.g. cd C:\Users\user\Miniconda3)
b. cd condabin
c. activate.bat
d. conda env create --file C:\pathto\odmlpython-cpu-mkl_win64.yml
e. conda env create - - file C:\pathto\odmlpython- cuda113_ win64.yml
- 87 -
Figure: Install Python environments from the YAML file on Windows
f. Conda will collect the packages it needs to install from the YAML file.
g. Then it will download and extract the packages, do the transaction and install
remaining packages with pip.
h. To make the Python environments completely operational with OpendTect

6.6's Machine Learning plugin it is necessary to get the cupti library on the sys-
tem. The cupti library is only needed for the CUDA 11 environment.
Cupti library
For Windows there are two ways to get the cupti library:
1. You can download cupti.lib and cupti64_ 2021.1.1.dll .

Then copy them into odmlpython-cuda113\Library\bin.
2. Or you can download and install CUDA Toolkit 11.3 update 1 from the Nvidia
CUDA site .
After installation add C:\Program Files\NVIDIA GPU Computing
Toolkit\CUDA\v11.3\extras\CUPTI\lib64 to the system or user environment
variable Path.
For information about how to use the custom environment in OpendTect 6.6 please
see the chapter Setup Python Settings.
Scipy missing DLL dependencies bug
- 88 -
On Windows you might experience a missing DLL bug in the scipy library. When this
happens the Machine Learning plugin will give an error message Inadequate
Python environment for Machine Learning plugin and the details will show a trace-
back. This traceback includes a FileNotFoundError: Could not find mod-
ule...gfortran-win_amd64.dll error.
Figure: Python environments scipy error when installed via YAML file
Scipy knows about this bug and the issue can be viewed on their GitHub page.
The following workaround has been tested and might also work for your Python
environment:
Set environment variable CONDA_DLL_SEARCH_MODIFICATION_ENABLE to 1
- 89 -
License server
Installing a license server on Linux

The FlexNet vendor package with the binaries that are needed can be downloaded
from the FlexNet Vendor Packages download page.
1. Download the Linux 64 (FlexNet v11.19.0.0) zip-file for OpendTect Pro & dGB Plugins
for OpendTect 6.4 and newer.
2. Create a directory on the server that is to be used as license server for OpendTect.
E.g. mkdir /opt/flexlm/opendtect
3. Unzip the zip-file in the directory that has been created in previous step.
4. The zip-file has the following files:

dgbld : The dGB vendor daemon
lmgrd : The FlexNet license manager daemon
lmutil : The FlexNet license management utilities
5. The next thing that is needed is the server license file. This license file will be provided
by the dGB support via [email protected].
Best is to place the license file in the same directory as lmgrd.
6. The license file needs to be edited. Hence, open the license file with a text editor like
Vim.
7. In the top of the license file there is the SERVER line. The SERVER line format is:
SERVER host hostid [port]
For example:
SERVER myserver 123456abcdef 27005
Note that if no port is specified then the FlexNet server will listen on the first available
TCP port in the range of 27000-27009. A valid port number is any unused port number
in the range of 1024 to 65535. Ports below 1024 are privileged port numbers.
Below the SERVER line is the DAEMON line. The DAEMON line format is:
DAEMON dgbld full_path_to/dgbld [PORT=port]
For example
DAEMON dgbld /opt/flexlm/opendtect/dgbld port=43454
- 90 -
It is important that you replace on the DAEMON line fullpath_to/dgbld with the path to
the dgbld file. In this example we will change it into: /opt/flexlm/opendtect/dgbld .
Note that if port is not specified then the vendor daemon will listen on an ephemeral ran-
dom port at run time. A valid port number is any unused port number in the range of
1024 to 65535. Ports below 1024 are privileged port numbers.
Save the license file.

8. Ensure that the proper firewall exceptions are in place. There should be two excep-
tions for:
1. The server TCP port, for the port specified in the license file, e.g. 27005
2. The vendor daemon TCP port, for the port specified in the license file, e.g. 43454
Starting a license server

The license server can be started by executing commands:
cd /opt/flexlm/opendtect
./lmgrd -c licensefile.lic -l dgbld_debug.log
Please refer to the debug.log file in case of issues.
To see the license server status you may run:

./lmutil lmstat - a
You will then see the port number the server uses, the path to the license file(s) that
are in use, whether the license server is up or down, how many licenses for the fea-
tures are issued and in use.
You may want to start the license server with crontab, e.g. add line (with crontab -e):
@reboot /opt/flexlm/opendtect/lmgrd - c /op-
t/flexlm/opendtect/licensefile.lic - l /op-
t/flexlm/opendtect/dgbld_debug.log
Error messages
You may encounter error messages when starting the license server. Listed here
are the most common ones and how to solve them.
l ./lmgrd: No such file or directory
- 91 -
This error occurs when either the file or the executable's loader is not present. To
find the executable's loader run the following command:
file lmgrd
It should show something like:
lmgrd: ELF 64- bit LSB executable, x86- 64, version 1
(SYSV), dynamically linked, interpreter /lib64/ld-lsb-x86-
64.so.3, for GNU/Linux 2.6.18, stripped
Now check whether the file /lib64/ld-lsb-x86-64.so.3 exists. If not then lsb needs
to be installed.
How to install lsb? This differs per distro:
l Debian/Ubuntu: sudo apt install lsb
l RHEL/CentOS/Rocky: yum install redhat-lsb
l OpenSUSE: zypper install lsb
l Cannot open "debug.log" as log file (- 1)
This error message occurs if the user has not enough permissions to create the
log file in the path specified. To solve this error message either become a user
that has permissions to create the log file in the specified path or create it in a
path where the user has sufficient permissions, e.g. in the user's home directory.
Debian/Ubuntu users may want to run sudo ./lmgrd - c license-
file.lic -l debug.log
l license manager: can't initialize: Cannot read license

file.
This error occurs if the user who starts the license server has no permissions to
read the license file. To solve this change the owner of the license file.
l Can't make directory /usr/tmp/.flexlm, errno: 2 (No such

file or directory)
This error message occurs if there is no /usr/tmp folder on the system. It can be
solved by creating a symbolic link /usr/tmp that points to /tmp, for example run as
root user:
ln -s /tmp /usr/tmp
- 92 -
l Error getting status: Cannot find license file (- 1,73:2
"No such file or directory")
This error message occurs when you run run ./lmutil lmstat -a, but there
is no license manager daemon running. It could be that the license manager dae-
mon failed to start. More information can be found in the debug.log file.
Client installation
Please see the installing floating licenses chapter for more information on how to
instruct the client on how to borrow the license from the server.
- 93 -
Installing a license server on Windows
The FlexNet vendor package with the binaries that are needed can be downloaded
from the FlexNet Vendor Packages download page.
1. Download the Windows 64 (FlexNet v11.19.0.0) zip-file for OpendTect Pro &
dGB Plugins for OpendTect 6.4 and newer.
2. Create a directory on the server that is to be used as license server for OpendTect.
E.g. create the directory C:\Program Files\FlexLM Licensing Server\OpendTect.
3. Unzip the zip-file in the directory that has been created in previous step.
4. The zip-file has the following files:

dgbld.exe : The dGB vendor daemon
lmgrd.exe : The FlexNet license manager daemon
lmtools.exe : The FlexNet LMTOOLS utility
lmutil.exe : The FlexNet license management utilities
5. Ideally the log file should be written into the directory containing the lmgrd executable.
Starting from FlexNet v11.17.2 this is possible on a path containing a space. If an older
FlexNet version is used as license server then we advise to write the log file into folder
C:\ProgramData\FLEXlm\lmgrd. Create this folder if it does not yet exist.
6. The next thing that is needed is the server license file. This license file will be provided
by the dGB support via [email protected].
Best is to place the license file in the same directory as lmgrd.exe.
7. The license file needs to be edited. Hence, open the license file with a text editor like
Notepad or Vim.
8. In the top of the license file there is the SERVER line. The SERVER line format is:
SERVER host hostid [port]
For example:
SERVER myserver 123456abcdef 27005
Note that if no port is specified then the FlexNet server will listen on the first available
TCP port in the range of 27000-27009. A valid port number is any unused port number
in the range of 1024 to 65535. Ports below 1024 are privileged port numbers.
Below the SERVER line is the DAEMON line. The DAEMON line format is:
DAEMON dgbld full_path_to/dgbld [PORT=port]
For example
DAEMON dgbld "C:\Program Files\FlexLM Licensing
- 94 -
Server\OpendTect\dgbld.exe" port=43454
It is important that you replace on the DAEMON line fullpath_to/dgbld with the path to
the dgbld.exe file. In this example we will change it into: "C:\Program Files\FlexLM
Licensing Server\OpendTect\dgbld.exe" . In case there are spaces in the path then the
path should be within double-quotes.
Note that if port is not specified then the vendor daemon will listen on an ephemeral ran-
dom port at run time. A valid port number is any unused port number in the range of
1024 to 65535. Ports below 1024 are privileged port numbers.
Save the license file.
9. Double-click the lmtools.exe file. This opens the LMTOOLS utility.
Figure: contents of the FlexLM Licensing Server\OpendTect directory
10. On Service/License File tab toggle on Configuration using Services and LMTOOLS

ignores license file path environment variables.
- 95 -
Figure: LMTOOLS utility: Service/License File tab
11. Click the Config Services tab.
1. At Service Name drop down create a new service name that will manage your
OpendTect Pro & dGB plugins licenses. The default service name is Flexlm Service
1. Best is to change this into a name like OpendTect-PRO License Manager.
2. At Path to the lmgrd.exe file browse to the directory that contains the lmgrd.exe file.
Select lmgrd.exe and then click button Open.
3. At Path to the license file browse to the directory that contains the license file. Select
the license file and then click button Open.
4. At Path to the debug log file put in the directory in which you want to write the dgbld_
debug.log file. Ideally C:\Program Files\FlexLM Licensing Server-
\OpendTect\dgbld_debug.log or C:\ProgramData\FLEXlm\lmgrd\dgbld_debug.log.
5. Toggle on Use Services and then toggle on Start Server at Power Up. This will
make sure that the license server will start automatically at power up or reboot.
- 96 -
Figure: LMTOOLS utility: Config Services tab
6. Click button Save Service and Yes in the confirmation window.
Figure: Save the service
7. LMTOOLS will show a warning window if the preferred path <Sys-

temDrive>\ProgramData to store service data is not set. You get this when dgbld_
debug.log is not configured in the ProgramData folder. You can ignore this when
there are enough permissions to read from and write to the dgbld_debug.log file.
Figure: Warning about Windows preferred path for service data
- 97 -
12. Ensure that the proper firewall exceptions are in place. There should be two excep-
tions (inbound application exceptions) for:
1. lmgrd.exe (TCP, for the port specified in the license file, e.g. 27005 )
2. dgbld.exe (TCP, for the port specified in the license file, e.g. 43454)
Starting a license server

1. Start LMTOOLS by double-clicking the lmtools.exe file.
2. Click the Start/Stop/Reread tab.
3. Make sure that the correct license server is selected in the list. Then click button Start
Server.
Figure: Starting dGB License Server
4. It is possible that the Windows firewall or your virus scanner blocks dgbld.exe . If this is
the case then you would want to add dgbld.exe to the list of Trusted Programs and/or
Firewall Rules. After making this change click button Start Server again.
5. If the license server start is successful LMTOOLS will show the message: Server Start
Successful.
- 98 -
6. Please refer to the debug.log file in case of issues.
This is an indication that the dgbld vendor daemon is not able to start:
(lmgrd) License server manager (lmgrd) startup failed:
(lmgrd) CreateProcess error code: 0x5 File= C:\Program
Files\FlexLM Licensing Server\OpendTect\dgbld.exe
License server manager (lmgrd) startup failed:
(lmgrd) CreateProcess error code: 0x5 File= dgbld.exe
Client installation
Please see the installing floating licenses chapter for more information on how to
instruct the client on how to borrow the license from the server.
- 99 -
Petrel* Plugin
Installing the PIP file
The plugin can be installed to Petrel* with the help of the PIP file in the usual way,
where you have to use the Plugin manager tool from the menu: File > Options > Plu-
gin manager. Press 'Install plugin' and select the downloaded PIP file. After install-
ation restart Petrel* to activate the plugin.
In order to install the updated version of the plugin, please remove the old one from
the Petrel* plugin manager. Then restart Petrel* and install the plugin as described
above.
The PIP files can be downloaded from the OpendTect Pro plugin for Petrel* down-
load page .
Figure: PetrelDirect download page
Choosing the preferred Communication Port

(TCP/IP)
- 100 -
By default the plugin should use the TCP/IP port 57375; in case this port is not avail-
able / accessible it can be changed in two ways:
l From plugin user interface, which is available under the 'Seismic interpretation' tab in
the Petrel* ribbon.
Figure: OpendTect Settings in ribbon
Figure: Settings for OpendTect connection
l By adding an environment variable DTECT_PETREL_PORT, and setting its value to

the preferred port number which is available for access.
PetrelDirect user documentation

For information about PetrelDirect go to the PetrelDirect chapter of the OpendTect
Pro and dGB Plugins Documentation.
Installation of PetrelDirect plugin in OpendTect

and Petrel*
PetrelDirect plugin allows OpendTect Pro users to seamlessly exchange data with
the Petrel* data store of a running Petrel* project. On the OpendTect side,
- 101 -
PetrelDirect plugin is installed as a part of OpendTect Pro. On the Petrel* side, Data
access for OpendTect (dGB) plugin must be installed. Currently supported Petrel*
versions are Petrel 2020*, Petrel 2021* and Petrel 2022*. This is a step-by-step
guide on how to install and configure PetrelDirect plugin in both OpendTect and
Petrel*.
OpendTect
1. OpendTect 6.6: PetrelDirect plugin is a part of OpendTect Pro:
2. Once OpendTect Pro is installed, PetrelDirect status button can be found in the lower
right corner of the main OpendTect window:
Figure: PetrelDirect status button
3. Click the button to check the Petrel connection settings. Make sure that the TCP port is
the same port as is set in Petrel*.
Figure: Petrel connection settings window
Petrel*
To be able to use PetrelDirect functionality in Opendtect Pro, Data access for
OpendTect (dGB) plugin must be installed in Petrel*. Installation can be done either
via Windows installer (MSI file) or Plugin Installer Package (PIP file).
- 102 -
Via Windows installer (MSI file):
The Windows installer does both the first-time plugin installation and an update of
already installed plugin to a newer version without any extra actions.
1. Download msi file from the OpendTect Pro plugin for Petrel* download page.
2. Run the msi file and follow the instructions.
3. Start Petrel*
4. Go to Seismic Interpretation tab and observe that OpendTect toolbar is there now.
Figure: OpendTect toolbar
Via Plugin Installer Package (PIP file):
For the plugin to update to a newer version, an old version must be uninstalled first:
1. Start Petrel*
2. In Petrel* main window: go to File > Options > Plugin Manager ...
3. In Ocean Plugin Manager window: uninstall the old version of Data access for
OpendTect (dGB) plugin by selecting it and clicking Uninstall button.
4. Close Petrel*
For the first-time installation or once the plugin version is uninstalled:
1. Download PIP file from the OpendTect Pro plugin for Petrel* download page.
2. Start Petrel*
- 103 -
3. In Petrel* main window: go to File > Options > Plugin Manager ...
Figure: File > Options > Plugin Manager
4. In Ocean Plugin Manager window:

1. Click on Install plugin button
2. Locate PIP file and click Open
3. Once the installation is finished click Close in both windows
5. Restart Petrel*
6. Go to Seismic Interpretation tab and observe that OpendTect toolbar is there now.
Figure: OpendTect toolbar
- 104 -
- 105 -
Troubleshooting
When experiencing issues with OpendTect there are several ways to troubleshoot:
You can
l look in the log file(s) inside the Survey Data Root\LogFiles and %TEMP% folder;
l review, modify and/or delete the OpendTect settings in the $HOME/.od folder on
Linux or %UserProfile%\.od on Windows;
l set the environment variable DTECT_DEBUG to yes to get more information

from OpendTect in the terminal or the Command Prompt when starting
OpendTect.
l set the environment variable QT_DEBUG_PLUGINS to 1 to get more information

about Qt in the terminal or the Command Prompt when starting OpendTect
(Linux and Windows) or for the OpendTect installer (Linux only).
l set the environment variable OSG_ NOTIFY_ LEVEL to DEBUG to get more
information about OSG in the terminal or the Command Prompt when starting
OpendTect.
l set the environment variable OD_ NOSCENE_ AT_ STARTUP to 1 to start

OpendTect without a scene. This is to determine whether there is a GPU driver
issue.
You can also
l look for an answer on the OpendTect Users Group or use your favorite search
engine;
l send an email to the OpendTect Users Group via [email protected]. To post

one needs to be a member. You can become a member by sending an email to
[email protected]. For more information please read the dgbe-
s.com Users Group Q&A page. Perhaps someone in the group experienced the
same issue and/or can tell you how to fix the issue;
- 106 -
l get priority support for commercial users by sending an email to the OpendTect
Support: [email protected];
l choose to submit a crash report in the event of a malfunction of OpendTect. This

crash report contains information that will help OpendTect's developers to locate
and fix the issue. If you choose to identify yourself in the comments, the crash
report will be added to your file by OpendTect developers, and OpendTect sup-
port may contact you for more information, or to inform you about the fixes of the
issue.
The crash report is a text-file that contains:
l Technical information about the crash.
l Technical information about the computer such as amount of memory, num-

ber of computer cores and a unique system ID. The unique system ID is
hashed, and cannot be used to identify the user.
l Any comments and/or email you add to the crash report before sending it.
l The IP address of the user.
Log files
Every time OpendTect starts a logfile is created in the Survey Data Root\LogFiles
folder. The naming of the log file is username_datetime.txt file.
Figure: log files in LogFiles folder
The OpendTect Installer writes logfiles into the /tmp folder on Linux or %TEMP%
folder on Windows, the naming of these files are:
l username_od_instmgr_install_log.txt
l username_od_instmgr_updcheck_log.txt
- 107 -
Figure: OpendTect installer log files
Settings files in .od folder

OpendTect writes its settings into the $HOME/.od folder on Linux or %User-
Profile%\.od on Windows. You can find the following files:
l qtsettings_odver; e.g. qtsettings_660
l settings
l settings_coltabs
l settings_dataroot
l settings_presentation
l settings_python
l settings_snapshot
l settings_welldisp
l settings_welltie
l survey
If one or more of the above listed settings files, like the settings_coltabs, does not
exist in your .od directory this means that OpendTect hadn't (yet) had any reason to
create that particular settings file.
Editing of these files is at your own risk. If you delete them OpendTect will create a
new file when needed. If a file gets corrupt then deleting is exactly what you should
do.
- 108 -
Debugging OpendTect
The default setting is that OpendTect only shows minimum debug info when starting
OpendTect via the start_ dtect script on Linux. Windows users will not see any
debug info when starting od_main.exe.
In order to get more debug information an environment variable DTECT_DEBUG

needs to be set to yes.
It is also possible to get debug information from Qt by setting an environment vari-

able QT_DEBUG_PLUGINS to 1. On Linux one can also get Qt debug information
when launching the OpendTect Installer.
When launching start_dtect from terminal on Linux the debug messages will be vis-
ible in the terminal.
Windows users will have to launch the Command Prompt, change the directory to
path_to\OpendTect\6.6.0\bin\win64\Release directory and from there run od_main_
console.exe
Figure: starting od_main_console.exe
- 109 -
Appendix A - FlexNet Licenses
Explained
Let us suppose you have received a 2-user license for a module. Normally you will
have obtained a floating license, meaning that you will be able to use the module
from any machine, and two users can be busy with it at the same time.
Server side: The License Manager Daemon
To be able to enforce the license, there must be a piece of software somewhere that
keeps track of who is using the module. That is the License Manager Daemon
(LMD). The LMD can run on any machine, also on machines that you will never use
the module itself on. Good candidates are stable UNIX servers.
When the LMD is started, it looks at a file containing the information about what
should be supported. This file is called the License File. Actually, the LMD can only
be started on the machine indicated in the license file. The license file could look like
this:
SERVER licserv 000347e8b845

DAEMON dgbld /apps/opendtect/6.6.0/bin/lux64/lm.dgb/dgbld
FEATURE dTect dgbld 6.600 1- jan- 2020 2 6592FDC619EA DUP_
GROUP=D
FEATURE dTectDS dgbld 6.600 1- jan- 2020 2 011D5153D870 DUP_
GROUP=D
The first line tells us that the LMD must be started on the machine licserv, with
FlexNet ID 000347e8b845. The second line is interesting for the LMD only but then
you see two actual license FEATURE lines (dTect and dTectDS). These licenses
are valid for versions 6.6 and lower, until the 1st of January 2020, for two users
(6.600 1-jan-2020 2).
The Client side: Your program
Now let's look at the machine that you run your software on. The program will at
some point in time need to check whether there is a license for what you are trying
to do. For this, the program looks at the same license file. It sees that it has to con-
tact the machine 'licserv' to ask for permission. The LMD keeps track of the number
- 110 -
of users already using the license feature. If a license is granted, your program will
go on, otherwise you'll get an error message.
Non-floating licenses
In some cases the software will never be used other than on a certain machine. In
that case a node-locked license can be issued. For such a license you do not need
to start a license manager daemon - an unlimited number of users can use the mod-
ule at the same time provided they work on that particular machine. A special case
is the unlimited demo license, which grants unlimited access for any number of
users on any machine. It goes without saying that this kind of license is always for
short periods.
Host identification
For the above schemes to work, the license server or the running machine must be
uniquely identified. Therefore, you will be asked to provide a unique host ID and a
hostname when you want to obtain a license. Different operating systems require dif-
ferent ways to obtain this information:
l Windows: On windows, there is a FlexNet utility that delivers both in a simple file that
can be sent by email. It is delivered together with commercial plugins so you need to
install one of those first. From the Start menu, select OpendTect > License Manager
Tools. Select the tab 'System Settings'. Then push the 'Save HOSTID Info to a File' but-
ton.
l Linux and Mac OS X: The hostname is obtained with the command hostname. The
host ID differs per distribution, but can always be obtained by the 'lmhostid' tool. This
tool is delivered with OpendTect, and can be run from the Utilities > Batch Programs
dialog.
If you need to obtain the host ID before OpendTect is installed:
l Windows: Open a 'command prompt' (For example, by running cmd) and issue a com-
mand like: ipconfig /all > c:\Temp\ipcfg.txt You can send ipcfg.txt or look
for the 'Physical address'.
l Linux and Mac OS X: Run ifconfig in a terminal. Or if ifconfig is not installed run
ip a. Look for HWAddr or ether. You want the MAC address, looking like
xx:xx:xx:xx:xx:xx - that's 6 groups of 2 hex numbers. Example: 00:1C:C0:38:22:F1.
Usually you would want to provide the one that is reported with devicename eth0 or
eno1. In doubt you can provide all.
- 111 -
If for some reason one of these commands does not work: we need the MAC
address of the main network card. Sending the IP address of a machine is never
helpful.
Conclusion (Manager's summary)
The FlexNet license system is based upon internet technology. Therefore you can
run your software on any machine, using any operating system, to get licenses from
any other machine regardless of operating system or physical location. Thus, a
Linux license server in Houston can manage the licenses for Windows, Linux and
Mac OS X machines in Houston, Caracas and Paris. The only restriction is the num-
ber of users actually using the 'feature' at a certain time, but that is what you pay for.
For more information, see the FlexNet Licensing End User Guide.
- 112 -

OpendTect Administrators Manual

Uploaded by

Copyright:

Available Formats

OpendTect Administrators Manual

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

OpendTect Administrators Manual

Uploaded by

Copyright:

Available Formats

OpendTect Administrator's

Copyright © 2002-2022 by dGB Beheer B.V.

For Machine Learning 7

For Machine Learning 9

For Machine Learning 12

Installing OpendTect on Windows 14

Installation via Installation Manager 14

Installing OpendTect on Linux 22

Installation via Installation Manager 24

Installing OpendTect on RHEL 6.x 31

Installing OpendTect on Mac OS X 33

Installation via Installation Manager 35

Licensing and Host IDs 41

Installing demo/node-locked licenses 44

Installing floating licenses 48

Setting the DGBLD_LICENSE_FILE environment variable 48

Using the FlexNet lmpath tool 49

Clear license installation 50

Prepare offline installation packages 52

Install OpendTect via offline installation package 54

Setup Python Settings 56

Setup a Custom Python environment 56

Setup Distributed Computing 58

Tips & tricks 61

Setup the preferred GPU with Windows settings 68

Setup NVIDIA GPU Performance Counters 76

Firewall Rules and Open Ports 80

Firewall rules For OpendTect 6.6 (Windows) 80

Information about the OpendTect Python environments 82

Installing Python environments with the OpendTect installer 85

Installing Python environments from YAML files 86

Download the OpendTect Python YAML files 86

Install the environments 86

Scipy missing DLL dependencies bug 88

Installing a license server on Linux 90

Starting a license server 91

Installing a license server on Windows 94

Starting a license server 98

Petrel* Plugin 100

Installing the PIP file 100

Choosing the preferred Communication Port (TCP/IP) 100

PetrelDirect user documentation 101

Installation of PetrelDirect plugin in OpendTect and Petrel* 101

Log files 107

Settings files in .od folder 108

Debugging OpendTect 109

Appendix A - FlexNet Licenses Explained 110

l GPU: Basic Intel graphics cards or Nvidia (e.g. recent GeForce/Quadro/NVS series)

l Memory: DDR4 memory, 16 GB of RAM, OpendTect itself needs at least 2 GB RAM.

l CPU: Intel/AMD processor with 64 bit support, 3+ GHz multi-core.

For Machine Learning

l GPU: Nvidia, GeForce or Quadro series.

l Ampere architecture cards (CUDA 11.1 and later)

l Nvidia A40 with 48 GB DDR6 memory and 10752 CUDA Cores

Please note that:

l For best performance OpenGL drivers should be up-to-date. For Machine Learning on

l Windows needs to be updated with the latest updates from Microsoft.

l RedHat/CentOS 7.2 and higher

l Ubuntu 16.04 and higher

l OpenSUSE Leap 15.1 and higher