Skip to content
This is a toolkit for launching and managing a graphistry stack on your servers.
Branch: master
Clone or download
Latest commit 0dcb625 May 8, 2019

Managing a Graphistry Deployment

Welcome to Graphistry! The graphistry-cli repository contains

  • Optional scripts to setup cloud Linux environment dependencies for Graphistry from scratch
  • Documentation for operating the Graphistry Docker container (install, configure, start/stop, & debug).

Quick Install

############ Environment
### Environment: Graphistry depends on nvidia-docker-2 and docker-compose
### Option 1 (10min): Sample environment configuration for Ubuntu 16.04 cloud environments:
git clone
bash graphistry-cli/ ubuntu-cuda9.2
### Option 2 (2min): AWS AMI `Graphistry-RHEL-20180801` on a G3.4  in Oregon region
ssh -i mykey
sudo service docker restart

############ Install & Launch
### Install
wget -O release.tar.gz "https://..."
tar -xvvf release.tar.gz
docker load -i containers.tar
docker-compose up

Quick Commands

Install docker load -i containers.tar Install the containers.tar Graphistry release from the current folder. You may need to first run tar -xvvf my-graphistry-release.tar.gz.
Start (interactive) docker-compose up Starts Graphistry, close with ctrl-c
Start (daemon) docker-compose up -d Starts Graphistry as background process
Stop docker-compose stop Stops Graphistry
Restart docker restart <CONTAINER>
Status docker-compose ps, docker ps, and docker status
API Key docker-compose exec central curl -s http://localhost:10000/api/internal/provision?text=MYUSERNAME Generates API key for a developer or notebook user
Logs docker logs <CONTAINER> (or docker exec -it <CONTAINER> followed by cd /var/log)


  • Instance & Environment Setup
    1. Prerequisites
    2. Instance Provisioning
    • AWS
    • Azure
    • On-Premises
    • Airgapped
    1. Linux Dependency Installation
    2. Graphistry Container Installation
    3. Start!
  • Configuration
  • Maintenance
    • OS Restarts
    • Upgrading
  • Testing
  • Troubleshooting

Instance & Environment Setup

1. Prerequisites

  • AWS Marketplace: Quota for GPU (P3.2+) in your region; ignore everything else below
  • Graphistry Docker container
  • Linux with nvidia-docker-2, docker-compose, and CUDA 9.2. Ubuntu 16.04 cloud users can use a Graphistry provided environment bootstrapping script.
  • NVidia GPU: K80 or later. Recommended G3+ on AWS and NC Series on Azure.
  • Browser with Chrome or Firefox

For further information, see Recommended Deployment Configurations: Client, Server Software, Server Hardware.

2. Instance Provisioning

AWS Marketplace

  • Use any of the recommended instance types (P3.2+)


  • Launch an official AWS Ubuntu 16.04 LTS AMI using a g3+or p* GPU instance.
  • Use S3AllAccess permissions, and override default parameters for: 200GB disk
  • Enable SSH/HTTP/HTTPS in the security groups
  • SSH as ubuntu@[your ami], centos@, or ec2-user@.

Proceed to the OS-specific instructions below.

For further information, see full AWS installation instructions.


  • Launch an Ubuntu 16.04 LTS Virtual Machine with an NC* GPU compute SKU, e.g., NC6 (hdd)
  • Check to make sure GPU is attached
$ lspci -vnn | grep VGA -A 12
0000:00:08.0 VGA compatible controller [0300]: Microsoft Corporation Hyper-V virtual VGA [1414:5353] (prog-if 00 [VGA controller])
	Flags: bus master, fast devsel, latency 0, IRQ 11
	Memory at f8000000 (32-bit, non-prefetchable) [size=64M]
	[virtual] Expansion ROM at 000c0000 [disabled] [size=128K]
	Kernel driver in use: hyperv_fb
	Kernel modules: hyperv_fb

5dc5:00:00.0 3D controller [0302]: NVIDIA Corporation GK210GL [Tesla K80] [10de:102d] (rev a1)
	Subsystem: NVIDIA Corporation GK210GL [Tesla K80] [10de:106c]
	Flags: bus master, fast devsel, latency 0, IRQ 24, NUMA node 0
	Memory at 21000000 (32-bit, non-prefetchable) [size=16M]
	Memory at 1000000000 (64-bit, prefetchable) [size=16G]
	Memory at 1400000000 (64-bit, prefetchable) [size=32M]

Proceed to the OS-specific instructions below.

For further information, see full Azure installation instructions.


See Recommended Deployment Configurations: Client, Server Software, Server Hardware.


Graphistry runs airgapped without any additional configuration. Pleae contact your systems representative for assistance with nvidia-docker-2 environment setup.

3. Linux Dependency Installation

If your environment already has nvidia-docker-2, docker, docker-compose, and CUDA 9.2, skip this section.

Ubuntu 16.04 LTS

    $ git clone
    $ bash graphistry-cli/ ubuntu-cuda9.2

RHEL 7.4 / CentOS 7

Note: Temporarily not supported on AWS/Azure

    $ sudo yum install -y git
    $ git clone 
    $ bash graphistry-cli/ rhel


Log off and back in (full restart not required): "$ exit", "$ exit"

Warning: Skipping this step means docker service may not be available

Warning: Skipping this step means Graphistry environment tests will not automatically run

Test environment

These tests run upon exiting the bootstrap. You can invoke them manually at any time:

    $ run-parts --regex "test*" graphistry-cli/graphistry/bootstrap/ubuntu-cuda9.2

Ensure tests pass for test-10 through test-40.

4. Graphistry Container Installation

Load the Graphistry containers into your system's registry:

docker load -i containers.tar

5. Start

Launch with docker-compose up, and stop with ctrl-c. To start as a background daemon, use docker-compose up -d.

Congratulations, you have installed Graphistry!

For a demo, try going to http://MY_SITE/graph/graph.html?dataset=Twitter, and compare to the public version.


Strongly Recommended:

After testing a base install works, configure the following:

  • Setup pivot password
  • Setup data persistence folders in case of restarts
  • Generate API Key for developers & notebook users

See for connectors (Splunk, ElasticSearch, ...), passwords, ontology (colors, icons, sizes), TLS/SSL/HTTPS, backups to disk, and more.


AWS Marketplace

See AWS Marketplace Administration

OS Restarts

Graphistry automatically restarts in case of errors. In case of manual restart or reboot:

  • On reboot, you may need to first run:
    • sudo systemctl start docker
  • If using daemons:
    • docker-compose restart
    • docker-compose stop and docker-compose start
  • Otherwise docker-compose up


  1. Backup any configuration and data: .env, docker-compose.yml, data/*, etc/ssl
  2. Stop the Graphistry server if it is running: docker-compose stop
  3. Load the new containers (e.g., docker load -i containers.tar)
  4. Edit and reload any config (docker-compose.yml, .env, data/*, etc/ssl)
  5. Restart Graphistry: docker-compose up (or docker-compose up -d)



If you downloaded the CLI:

run-parts --regex "test*" graphistry-cli/graphistry/bootstrap/ubuntu-cuda9.2

Note that these are not deep tests of the environment.


  • Installation repositories are accessible:
  • Nvidia infrastructure setup correctly
    • nvidia-smi reports available GPUs
    • nvidia-docker run nvidia/cuda nvidia-smi reports available GPUs
    • nvidia-docker run graphistry/cljs:1.1 npm test reports success (see airgapped alternative as well)
    • Using the image listed in docker images, running nvidia-docker run nvidia-smi reports available GPUs
  • Configurations were generated:
    • .config/graphistry/config.json
    • httpd-config.json
    • pivot-config.json
    • viz-app-config.json
  • Services are running: docker ps reveals no restart loops on:
    • monolith-network-nginx
    • monolith-network-pivot
    • monolith-network-viz
    • monolith-network-mongo
    • monolith-network-db-bu
    • monolith-network-pg
  • Services pass initial healthchecks:
  • Pages load
    • shows Graphistry homepage
    • clusters and renders a graph
    • loads a list of investigations
    • loads a list of connectors
    • ^^^ When clicking the Status button for each connector, it reports green
    • Opening and running an investigation in uploads and shows a graph
  • Data uploads


Create the below notebook, fill in appropriate values for GRAPHISTRY. The expected result is a link, that when you click it, shows a graph with 3 nodes.

    'server': '', #no http, just domain
    'protocol': 'http',
    'key':  'MY_API_KEY'

!pip install pandas
import pandas as pd
edges_df=pd.DataFrame({'src': [0,1,2], 'dest': [1,2,0]})

!pip install graphistry
import graphistry

graphistry.bind(source='src', destination='dest').edges(edges_df).plot(render=False)

For further information about the Notebook client, see the OSS project PyGraphistry ( PyPI ).


See further documentation.

You can’t perform that action at this time.