SID Visualization

AI & Data Science Cybersecurity
1

Hi everyone, I successfully ran the SID Visualization example following the steps in the official link:

The docker container was built without any problem and the run.py script runs okay:

✔ gui Built 0.0s
✔ Container sid_visualization-triton-1 Running 0.0s
✔ Container sid_visualization-gui-1 Started 24.3s
✔ Container sid_visualization-morpheus-1 Running

python examples/sid_visualization/run.py
–debug
–triton_server_url=triton:8000
–input_file=./examples/data/sid_visualization/group1-benign-2nodes.jsonlines
–input_file=./examples/data/sid_visualization/group2-benign-50nodes.jsonlines
–input_file=./examples/data/sid_visualization/group3-si-50nodes.jsonlines
–input_file=./examples/data/sid_visualization/group4-benign-49nodes.jsonlines

└─ morpheus.ControlMessage → morpheus.ControlMessage
Websocket server listening at: ‘0.0.0.0:8765’
====Building Segment Complete!====
\[Completed\]Inference rate\[ # \] 152160 messages in 00m:38s, Throughput: 3904.71 messages/s

However, I don’t know how to consume the green graph with nodes showed in the link:

(Image for reference below)

image
image1018×630 134 KB

Any idea how to manage this?

2

Hi @fmorales4 -
Could you share where you are deploying the Morpheus AND sid_visualization-gui container?

The GUI uses Electron so you A) need to ensure the deployed system is not running a headless Ubuntu, for example and B) you have a method to access that GUI from a remote machine.

If you are deploying everything on a local machine the GUI should launch. If not we can follow up.

4

Thanks for the response. Both Docker containers (morpheus:sid-viz and nvcr.io/nvidia/morpheus/morpheus-tritonserver-models:25.06) are indeed running on a headless Ubuntu server, not on my local machine. I only have access to the NVIDIA GPU through a remote server.

Is it possible to run the pipeline on the remote server while using my local computer solely for visualization? I’d really appreciate any insights on how to achieve this.

5

That should be possible but not something we’ve tested in this example. Your best bet is to configure the MORPHEUS_SOCKET_URL in the example docker config file to an open websocket url.

Then run the visualization independently on your machine via morpheus-visualizations/GraphVis at branch-25.10 · nv-morpheus/morpheus-visualizations · GitHub

6

Thanks @aenemark I’ll give it a try, I appreciate your response.