Skip to content

Latest commit

 

History

History
147 lines (96 loc) · 5.56 KB

setup.md

File metadata and controls

147 lines (96 loc) · 5.56 KB
Raw

Setup graph-app-kit

The below provides quick launchers for AWS-based deployments. See manual setup for alternative instructions, and check for database-specific integrations such as for Amazon Neptune.

1. Launch graph-app-kit

Option 1 - Full (Recommended):

Launch Stack

  • If AWS reports Please select another region, use the Select a Region dropdown in the top right menu

Launches:

  • GPU instance
  • Web-based live editing
  • Core toolkit: Graphistry, public + private Streamlit dashboards, Jupyter notebooks, RAPIDS.ai Python GPU ecosystem

Tenants launching GPUs for the first time may need to request 4+ vCPU of g4, p3, or p4 capacity

Option 2 - Minimal:

Steps:

  • Get a free or self-managed Graphistry server account with username+pass
  • Launch Stack
    • If AWS reports Please select another region, use the Select a Region dropdown in the top right menu.

Launches:

  • AWS CPU-mode instance + hub.graphistry.com account
  • Edit dashboard views from your terminal
  • Included: Public Streamlit dashboards linked against a remote Graphistry account
  • Not included: Local Graphistry (GPU), private dashboards, Jupyter, RAPIDS.ai (GPU)
  1. Launch configuration: Details parameters
  • Set stack name to anything, such as graph-app-kit-a
  • Set VPC to one that is web-accessible
  • Set Subnet to a web-accessible subnet in the VPC ("subnet-...")
  • Set GraphAppKitKeyPair to any where you have the SSH private.key

If using the minimal template, fill in details for your Graphistry account

  1. (Optional): Monitor instance launch for progress and errors

  • Click the Resources tab and follow the link to the EC2 instance AWS console page after it gets generated

  • Click on the instance to find its public IP address

  • Login and watch:

ssh -i /my/private.key [email protected] 
### ssh -i /my/private.key [email protected] for Minimal launcher

tail -f /var/log/cloud-init-output.log -n 1000

3. Graph!

Go to your public Streamlit dashboard and start exploring: http://[the.public.ip.address]/public/dash

Login

  • Upon launch completion, you will have a full suite of graph tools located at http://[the.public.ip.address]

  • Web login using credentials admin / i-theInstanceID

  • SSH using the instructions from step 2

  • Note: The minimal launcher has no web admin portal, just SSH and Streamlit

URLs for full stack

  • Graphistry: GPU-accelerated visual analytics + account login
    • http://[the.public.ip.address]
    • Login as admin / your-aws-instance-id
    • Installed at /home/ubuntu/graphistry
    • You can change your admin password using the web UI
  • Streamlit: Public dashboards
    • http://[the.public.ip.address]/public/dash
    • Installed at /home/ubuntu/graph-app-kit/public/graph-app-kit
    • Run as src/docker $ docker-compose -p pub run -d --name streamlit-pub streamlit
  • Streamlit: Private dashboards
    • http://[the.public.ip.address]/private/dash
    • Installed at /home/ubuntu/graph-app-kit/private/graph-app-kit
    • Run as src/docker $ docker-compose -p priv run -d --name streamlit-priv streamlit
  • Jupyter: Data science notebooks + Streamlit dashboard live-editing
    • http://[the.public.ip.address]/notebook
    • Live-edit graph-app-kit view folders notebook/graph-app-kit/[public,private]/views

URLs for minimal stack

  • Streamlit: Public dashboards
    • http://[the.public.ip.address]/public/dash
    • Installed at /home/ubuntu/graph-app-kit/public/graph-app-kit
    • Run as src/docker $ docker-compose up -d streamlit

4. Optional - administer

Advanced users can SSH into the server to manipulate individual services:

System visibility

# launch logs
tail -f /var/log/cloud-init-output.log -n 1000

# app logs
sudo docker ps
sudo docker logs -f -t --tail=1 MY_CONTAINER

# stats
sudo htop
sudo iftop
top

Graphistry

For more advanced Graphistry administration, so the Graphistry admin docs repo

# restart a graphistry container
cd graphistry && sudo docker-compose restart MY_CONTAINER  # do *not* run it's caddy (v2)

# restart caddy (Caddy 1 override over Graphistry's Caddy 2)
cd graphistry && sudo docker-compose -f docker-compose.gak.graphistry.yml up -d caddy

Streamlit

Use docker-compose project names (-p the_name) to distinguish your public vs private dashboards:

cd graph-app-kit/public/graph-app-kit && docker-compose -p pub run -d --name streamlit-pub streamlit
cd graph-app-kit/private/graph-app-kit && docker-compose -p priv run -d --name streamlit-priv streamlit

5. Next steps

Continue to the instructions for creating custom views and adding common extensions like TLS, public/private dashboards, and more