Skip to content

JURI Installation Instructions

The Jülich Reporting Interface (JURI) serves as the web presentation portal for LLview. It renders job metrics, user tables, and detailed visual reports.


Dependencies

Client-Side (Browsers)

JURI runs in modern web browsers and is tested against recent versions of Firefox, Google Chrome, and Safari.

JURI bundles the following open-source client libraries: - AG-Grid Community - Bootstrap - d3.js - Font Awesome (v4) - Handlebars - jQuery - Mermaid - Plotly.js - svg-pan-zoom

Web Server Requirements

LLview's role-based access control (RBAC) relies on Apache access controls:

  • Apache HTTP Server (>= 2.4)
  • Enabled Modules:
    • mod_authz_groupfile
    • mod_headers
    • mod_rewrite
  • Directive Requirement:
  • PHP (>= 7.4)

Configuration

1. Data Staging from LLview Server

LLview Server regularly pushes reports and metrics to the Web Server into a target staging directory (referred to here as $LLVIEW_WEB_DATA):

  1. Create the data destination folder on the Web Server:

    mkdir -p $LLVIEW_WEB_DATA
    

  2. Authorize file transfers from the LLview Server:
    JURI bundles the restricted rrsync.pl utility under $JURI_HOME/utils/. Append the public key generated on the LLview Server (see Server setup) to ~/.ssh/authorized_keys on the Web Server:

    from="<LLVIEW_SERVER_IP>",command="$JURI_HOME/utils/rrsync.pl -wo $LLVIEW_WEB_DATA",no-agent-forwarding,no-port-forwarding,no-pty,no-user-rc,no-X11-forwarding <PUBLIC_KEY_CONTENT>
    

2. User Roles and Group Permissions

LLview enforces permissions across multiple roles (User, Principal Investigator, Project Admin, Mentor, and Support).

Group definitions are written to $LLVIEW_WEB_DATA/data/sec_files/.htgroups, which is generated on the LLview Server and synced across. The generated .htaccess files point to this .htgroups file.

Important: Because .htaccess requires absolute paths (or paths relative to DocumentRoot), the exact location of $LLVIEW_WEB_DATA must be configured in ~/.llview_server_rc on the LLview Server.

Ensure your Apache virtual host enables override controls:

<Directory "/path/to/web/data">
    AllowOverride All
</Directory>

3. .htaccess and Authentication

Authentication is managed via a root .htaccess file placed in $LLVIEW_WEB_DATA. A template is provided in the JURI repository with commented blocks for: - .htpasswd (basic auth) - LDAP - OpenID Connect (OIDC / mod_auth_openidc)

Customizing login.php

The landing page (login.php) reads the authenticated user identity (via $_SERVER['REMOTE_USER'] or $_SERVER['PHP_AUTH_USER']) and matches it with the permission map generated by the webservice step of the server collect action. If your institutional login environment exposes user IDs under a different server variable, adapt login.php accordingly.

Compressed File Delivery (.gz)

To optimize bandwidth, HTML reports can be transferred pre-compressed. The top-level .htaccess file includes rewrite rules to serve .html.gz files transparently with the proper Content-Encoding: gzip header.

Tip (Quick Access to Job Reports):
Supporters can jump directly to reports using the Jump to JobID field. To enable clean direct URLs (e.g., https://llview.your-domain.org/<JobID>), place an .htaccess file in the parent folder of $LLVIEW_WEB_DATA with the following rewrite rule:

RedirectMatch "^/(.*?)/([0-9]+)$" "/$1/login.php?jobid=$2"


Installation Steps

1. Clone the JURI Repository

Clone JURI into your chosen location on the Web Server:

git clone https://github.com/FZJ-JSC/JURI.git
This installation path represents $JURI_HOME.

JURI uses symbolic links to separate web code ($JURI_HOME) from dynamically synced metrics ($LLVIEW_WEB_DATA).

Run the helper script to create the necessary symlinks:

$JURI_HOME/utils/linkjuri.sh "$JURI_HOME" "$LLVIEW_WEB_DATA"

After running linkjuri.sh, $LLVIEW_WEB_DATA will contain: - The data/ directory (populated by LLview Server) - Symbolic links to JURI's HTML, PHP, and JavaScript assets

Warning: The top-level .htaccess file is not linked automatically by linkjuri.sh to prevent overwriting site-specific security configurations. Copy the template manually:

cp $JURI_HOME/.htaccess $LLVIEW_WEB_DATA/.htaccess

3. Publish to DocumentRoot

Expose the combined directory to your web server by linking it into Apache's DocumentRoot:

ln -s $LLVIEW_WEB_DATA /srv/www/htdocs/<cluster_name>


Optional: Enabling LLview-AI Features

LLview includes optional AI capabilities: an interactive AI Chatbot and AI-Generated Job Summaries. These can be enabled in $JURI_HOME/ui-config.json.

Configuration (ui-config.json)

{
  "llviewai": {
    "enableChatBot": false,
    "chatBotEndpoint": "",
    "chatBotSupportEmail": "",
    "chatBotLLMProvider": {
      "name": "",
      "url": ""
    },
    "enableAIReports": false,
    "aiReportEndpoint": "",
    "aiReportSupportEmail": "",
    "aiReportLLMProvider": {
      "name": "",
      "url": ""
    },
    "enableTableAttachment": false
  }
}

Field Descriptions:

  • Chatbot Settings:
  • enableChatBot: Enable or disable the interactive chat feature (true/false).
  • chatBotEndpoint: (Required if enabled) API endpoint URL for the chat service.
  • chatBotSupportEmail: (Optional) Support email displayed in the disclaimer.
  • chatBotLLMProvider: (Optional) Provider name and url shown in the disclaimer.
  • enableTableAttachment: (Optional) Allows users to attach the currently visible data table as context for the chatbot (true/false).
  • AI Report Settings:
  • enableAIReports: Enable or disable AI-generated job report summaries (true/false).
  • aiReportEndpoint: (Required if enabled) API endpoint URL for report generation.
  • aiReportSupportEmail: (Optional) Support email displayed in the disclaimer.
  • aiReportLLMProvider: (Optional) Provider name and url shown in the disclaimer.

Backend Endpoint Contract

Before enabling these toggles, ensure your backend API endpoints exist and handle POST requests as expected:

AI Job Reports Endpoint: Accepts multipart/form-data containing the report PDF and user prompt:

const pdfResp = await fetch("/path/to/jobreport.pdf");
if (!pdfResp.ok) { 
    throw new Error(`Could not fetch PDF (${pdfResp.status})`); 
}
const pdfBlob = await pdfResp.blob();
const file = new File([pdfBlob], "jobreport.pdf", { type: "application/pdf" });

const form = new FormData();
form.append("file", file);
form.append("userprompt", "Summarize performance bottlenecks");

const resp = await fetch("/api/aireport", {
    method: "POST",
    headers: { "Accept": "application/json" },
    body: form
});

AI Chatbot Endpoint: Accepts a JSON payload containing the conversation messages:

const message = "Why did job 12345 fail?";

const resp = await fetch("/api/chat", {
    method: "POST",
    headers: { 
        "Content-Type": "application/json",
        "Accept": "application/json" 
    },
    body: JSON.stringify({ 
        messages: [{ role: "user", content: message }] 
    })
});