Ecosynther Now Available as AWS Image

Hello everyone, I am pleased to announce we now have a publicly available Amazon EC2 (cloud computing) AWS image of our Ecosynther software - you can run this online . I have written out instructions and placed them here. In case there are any issues accessing that document, I've also copied the instructions below. Please reach out to us if there are issues implementing what I've put together.

The steps listed below are for producing your own mirrored instance of Ecosynther

  1. Register for AWS account (unless you already have one).
  2. Navigate to https://console.aws.amazon.com/ec2/
    • This will redirect to your region, but please note that the Ecosynther image is located on us-west-2 (Oregon) in case it does not appear in your search
  3. Click "Launch Instance" button under the "Instances" menu
  4. Select "Community AMIs" from the left menu
  5. Search for "Ecosynther", optionally checking Ubuntu under "Operating system"
  6. Click the select button
    • IMPORTANT: Under filter by, select "GPU Instances". Ecosynther - as configured in this AMI - will not run properly on an instance without a GPU.
  7. At the time these instructions were written, only one Ubuntu GPU enabled instance was on offer. In case that changes, the type should be listed as something similar to "g2.2xlarge".
  8. Click "Next: Configure Instance Details". Leave as default.
  9. Click "Next: Add Storage". Verify drive size is set to 250GB. Check your data set size, increase if necessary.
  10. Click "Review and Launch". You can configure more if you want, but it is not necessary for our setup. Verify configuration and hit "Launch". Please note this instance cannot be run on AWS's free usage tier. Please keep this is mind as you plan ahead for your data usage.
  11. The dashboard will provide with a .pem key file and instructions for downloading, etc. Do not lose/delete your .pem file, you only get the one! Log into your console. Select "Instances". Click your mirrored copy of Ecosynther and click the "Connect" button at the top. This will provide you a command that looks like: ssh -i ILoveTrees.pem ubuntu@54.148.7.117. Note, if it says the user is ec2-user, change that to ubuntu as ec2-user is for a different linux variant.
  12. From your preferred command line shell, navigate to the directory with your .pem file. Enter the ssh command mentioned in the step above. You can create additional users that do not require a .pem file within your instance if you so desire.
  13. Once inside, you will find the project root directory at /home/ubuntu/Eco7/EcosyntherFull.
    • The software comes preinstalled on the image, but should something go awry we use make to compile the program. Run "make clean" and then "make" from this directory.
  14. From <ecosynther root>/bundler-v0.4-source, please run sudo ./xWindowlessPrep.sh. This will ensure that the X windowing system is properly configured to run on AWS.
  15. Load your dataset onto your instance using wget, WinSCP, or similar tools. From the directory with your unpacked image files, please run sudo /full/path/to/Ecosynther/RunEcosyntherFull.sh
    • Your resulting data set will be written to <your image directory>/pmvs/models
  16. In case you have previous experience with Ecosynther, here are some thing you will want to keep in mind should errors appear or you wish to tinker with the code.
    • Please ensure the libraries you use are 64-bit.
    • Please ensure that your camera's CCD value is listed under <ecosynther root>/bundler-v0.4-source/bin/extract_focal.pl
    • Since AWS is a headless server (read: no monitor), some additional configuration has to be performed on top of code you may have seen before. This necessitated the script written under xWindowlessPrep.sh. After that script is run, the DISPLAY value will be reset as part of the normal ecosynther process.
  17. Try it out!  Give us feedback on our blogs/Forum!

Views: 568

Comment

You need to be a member of Ecosynth to add comments!

Join Ecosynth

Members

© 2017   Created by Erle Ellis. Content is made available under CC BY 4.0.   Powered by

Badges  |  Report an Issue  |  Terms of Service