Settings
Settings allow you to configure global options for LLMOS.
note
Only admin users can manage these settings.
General Settings
auth-token-max-ttl-minutes
Type | Default | Description |
---|---|---|
Integer | 129600 (30 days) | Maximum TTL (in minutes) for all authentication tokens. When set to 0, tokens will never expire. |
auth-user-session-max-ttl-minutes
Type | Default | Description |
---|---|---|
Integer | 720 (12 hours) | Custom TTL (in minutes) for a user authentication session. |
default-notebook-images
Type | Default | Description |
---|---|---|
JSON String | See below JSON string | Default list of images to use for notebooks. JSON keys represent image tags, and values are lists of notebook images. |
{
"code-server": [
{
"containerImage": "ghcr.io/oneblock-ai/codeserver-python:v0.1.0",
"description": "Visual Studio Code + Conda Python",
"default": true
}
],
"jupyter": [
{
"containerImage": "ghcr.io/oneblock-ai/jupyter-scipy:v0.1.0",
"description": "JupyterLab + PyTorch",
"default": true
},
{
"containerImage": "ghcr.io/oneblock-ai/jupyter-pytorch:v0.1.0",
"description": "JupyterLab + PyTorch"
},
{
"containerImage": "ghcr.io/oneblock-ai/jupyter-pytorch-full:v0.1.0",
"description": "JupyterLab + PyTorch + Common Packages"
},
{
"containerImage": "ghcr.io/oneblock-ai/jupyter-pytorch-cuda:v0.1.0",
"description": "JupyterLab + PyTorch + CUDA"
},
{
"containerImage": "ghcr.io/oneblock-ai/jupyter-pytorch-cuda-full:v0.1.0",
"description": "JupyterLab + PyTorch + CUDA + Common Packages"
},
{
"containerImage": "ghcr.io/oneblock-ai/jupyter-tensorflow:v0.1.0",
"description": "JupyterLab + PyTorch"
},
{
"containerImage": "ghcr.io/oneblock-ai/jupyter-tensorflow-full:v0.1.0",
"description": "JupyterLab + PyTorch + Common Packages"
},
{
"containerImage": "ghcr.io/oneblock-ai/jupyter-tensorflow-cuda:v0.1.0",
"description": "JupyterLab + PyTorch + CUDA"
},
{
"containerImage": "ghcr.io/oneblock-ai/jupyter-tensorflow-cuda-full:v0.1.0",
"description": "JupyterLab + PyTorch + CUDA + Common Packages"
},
{
"containerImage": "ghcr.io/oneblock-ai/jupyter-pipeline:v0.1.0",
"description": "JupyterLab + Elyra Pipeline"
}
],
"rstudio": [
{
"containerImage": "ghcr.io/oneblock-ai/rstudio-tidyverse:v0.1.0",
"description": "RStudio + Tidyverse",
"default": true
}
]
}
global-system-image-registry
Type | Default | Description |
---|---|---|
String | "" | Default registry for LLMOS operator & system images (e.g., myregistry.com:5000). |
log-level
Type | Default | Description |
---|---|---|
String | info | Defines the log level for the LLMOS API server. |
model-service-default-image
Type | Default | Description |
---|---|---|
String | "" | Default serving image for the LLMOS modelService. |
server-url
Type | Default | Description |
---|---|---|
String | "" | Default Dashboard installation URL. Must use HTTPS. All nodes in your cluster must be able to reach this URL. |
upgrade-check-enabled
Type | Default | Description |
---|---|---|
Bool | true | Enables or disables automatic upgrade checks. |
upgrade-check-url
Type | Default | Description |
---|---|---|
String | https://llmos-upgrade.1block.ai/v1/versions | Default LLMOS upgrade check URL. Only used when upgrade-check-enabled is set to true. |
UI Settings
ui-index
Type | Default | Description |
---|---|---|
String | https://releases.1block.ai/dashboard/latest/index.html | HTML index location for the LLMOS Dashboard UI. |
ui-pl
Type | Default | Description |
---|---|---|
String | "llmos" | Specifies the private-label company name. |
ui-source
Type | Default | Description |
---|---|---|
Enum | "Dynamic" | Controls whether UI assets are served locally by the API server or fetched from the remote URL specified in ui-index . |