Skip to content

Latest commit

 

History

History
122 lines (74 loc) · 5.54 KB

CF.md

File metadata and controls

122 lines (74 loc) · 5.54 KB

Virtual server in the cloud with access to Earth Observation (EO) data

In addition to the standard CDSE services the hackathon participants can launch virtual server in the CREODIAS platform provided by CloudFerro.

For the access to the CREODIAS platform each team will receive one Keystone credential and an SSH key to be used with the pre-configured OSGeoLive server with additional services (Rstudio, JupyterHub, Shiny, Apache Superset).

During the hackathon in the Horizon Dashboard the participants have to use the option Keystone credentials and use the Domain cloud_078898 and region WAW3-2.

Horizon dashboard login screen

After the first login, the participants should be able to see the pre-launched server with GPU support. The participants from the Dashboard can terminate the preconfigured server and can launch additional service till the allowed overall limit per team, which are:

  • max 20 instances
  • max 20 CPU cores total in all instances
  • max 50 GB RAM total in all instances
  • max 10 disks with a max 1000 GB storage in total

From the virtual servers you can access the CDSE resources and the additional datasets which are mentioned in the Data Catalogue.

:::{Note} The sample notebooks to access the datasets together with some description of the data are available in these folders in this repository:

Step by step instructions to recreate the customized image with additional services

  1. After login in to the Horizon dashboard go to the Instances in the left menu and choose Launch Instance.

Horizon dashboard instances

  1. In the new window first you give a name to your instance

Instance name

  1. Then select the source of the image: OSGeoLive 16.0

Image source

  1. The next step is to select the flavor of the instance e.g. size of CPU, RAM and disk.

:::{Important} The instance should have at least 40GB of disk! :::

Flavor

  1. Under the networks add both networks

Networks

  1. Under the Security Groups enable the following ports for incoming connections:
  • 22 (SSH)
  • 3389 (RDP)
  • 3838 (Shiny)
  • 8000 (JupyterHub)
  • 8088 (Superset)
  • 8787 (Rstudio)

:::{Note} In the account there can be already predefined security groups contain these port. If not, you have to create it following these instructions :::

Security Groups

  1. Finally you select or create an SSH key pair that will be used to connect to the instance for the first time.

SSH key

  1. Launch the instance and wait until it will be running

  2. Create and associate an external IP following these steps

Add IP

  1. Then you can access your instance with an SSH client, and the first step will be to create a password for the default eouser. After creating the password the connection will be automatically closed.

Initialize password

  1. To enable RDP and add the additional services you have to run the following shell script: add_services_nogpu.sh For that, you have to upload/copy (e.g. WinSCP) to the server. Make it executable and run in a terminal.

    chmod +x add_services_nogpu.sh
./add_services_nogpu.sh

    :::{Important} Before you run the script replace in the script on row 136 the Superset admin password <your_superset_password> with your own password. :::

    :::{Note} The script runs and install all the services and libraries. The script restarts several times the server during the installation process and can last up to 1 hour. You can check the installation status with more ~/.service_install_status in the terminal. When it finishes the status will be everything went well.
    Instal finished :::

Accessing the services

RDP

The service is available under the external IP:3389 created under point 9 above in standard RDP clients. In case you are behind proxy you can tunnel the internal IP:3389 through SSH to your localhost and connect to your localhost port with the RDP client.

Shiny

The service is available in a browser under the external IP:3838 created under point 9 above.

JupyterHub

The service is available in a browser under the external IP:8000 created under point 9 above. The username is eouser and password what is defined under point 10.

Apache Superset

The service is available in a browser under the external IP:8088 created under point 9 above. The username is eouser and password what is defined under point 11 defined in the add_services_nogpu.sh.

Rstudio

The service is available in a browser under the external IP:8787 created under point 9 above. The username is eouser and password what is defined under point 10.