Skip to content

Latest commit

 

History

History
68 lines (61 loc) · 3.67 KB

serverSetup.md

File metadata and controls

68 lines (61 loc) · 3.67 KB

GAMMA-server Google Cloud Setup

  • Create a new project
  • Go to Compute Engine and to VM instances
  • Create Instance
    • Region: Something close-by: e.g europe-west3, zone: e.g a
    • Machine Family: General Purpose
    • Series: E2
    • Machine type: e2-medium (2 vCPU, 4GB memory)
    • Boot Disk:
      • Operating System: Ubuntu
      • Version: Ubuntu 22.04 LTS
      • Balanced persistent disk
      • Size 20 GB (At least 16 GB are needed)
    • Firewall: Allow HTTP and HTTPS traffic
    • Advanced options: Networking: Add a custom tag. e.g ar-population-instance
  • Go to Set up firewall rules: Create Firewall Rule
    • Direction Traffic: Ingress
    • Action on match: Allow
    • Targets: Specified target tags
    • Target tags: Same custom tag. e.g ar-population-instance
    • Source IPv4 ranges: 0.0.0.0/0
    • Check the TCP box
    • Write 8080 in the Ports field
  • Go to VM instances
  • Log in to the vm instance with e.g the browser ssh window or with the gcloud command from the local terminal
  • Install miniconda:
    • wget https://repo.anaconda.com/miniconda/Miniconda3-py38_4.12.0-Linux-x86_64.sh
    • bash Miniconda3-py38_4.12.0-Linux-x86_64.sh
  • Clone the Gamma-server repo
  • Create conda environment: cd GAMMA-server && conda env create -f env.yml
  • conda activate gamma
  • Copy the checkpoints and extern dependencies to the repo: wget https://polybox.ethz.ch/index.php/s/FtM3i343RqdDldv/download -O checkpoints.zip
  • Unzip them such that results and extern are in the root directory of the repo (GAMMA-server/extern).
  • run python gamma_server.py
  • To test if everything works, use curl -X POST -H "Content-Type: application/json" -d '[1, 2, 3, 2, 3, 4]' http://34.159.11.198:8080 where the ip is the external ip from the VM displayed under VM instances. The server should start printing and respond with a json.
  • Troubleshoot: If when executing the server an error with conversions occurs try changing the following lines to look like this in the file /home/--username--/miniconda3/envs/gamma/lib/python3.8/site-packages/torchgeometry/core/conversions.py:
mask_c0 = mask_d2 * mask_d0_d1
mask_c1 = mask_d2 * ~(mask_d0_d1)
mask_c2 = ~(mask_d2) * mask_d0_nd1
mask_c3 = ~(mask_d2) * ~(mask_d0_nd1)
  • Troubleshoot: If there is no reaction from the server, the network settings of the VM might be the problem. Try in a new VM terminal to CURL the local ip address that is printed by the server. If the server reacts then, check the firewall settings of the VM.

GAMMA-server local setup

If you have a e.g. desktop and use it as the server, you only need to do the following steps above:

  • Install miniconda:

    • wget https://repo.anaconda.com/miniconda/Miniconda3-py38_4.12.0-Linux-x86_64.sh
    • bash Miniconda3-py38_4.12.0-Linux-x86_64.sh

  • Clone the Gamma-server repo

  • Create conda environment: cd GAMMA-server && conda env create -f env.yml

  • conda activate gamma

  • Copy the checkpoints and extern dependencies to the repo: wget https://polybox.ethz.ch/index.php/s/FtM3i343RqdDldv/download -O checkpoints.zip

  • Unzip them such that results and extern are in the root directory of the repo (GAMMA-server/extern).

  • run python gamma_server.py

  • Troubleshoot: If when executing the server an error with conversions occurs try changing the following lines to look like this in the file /home/--username--/miniconda3/envs/gamma/lib/python3.8/site-packages/torchgeometry/core/conversions.py:

mask_c0 = mask_d2 * mask_d0_d1
mask_c1 = mask_d2 * ~(mask_d0_d1)
mask_c2 = ~(mask_d2) * mask_d0_nd1
mask_c3 = ~(mask_d2) * ~(mask_d0_nd1)