​

Using the Dashboard: Complete Guide

The Dashboard is your central command center - providing real-time system health, active incidents, agent status, infrastructure overview, and live activity streams all in one view.

---

​

Dashboard Overview

The dashboard header shows:

• Title: “Dashboard” with description “System overview and real-time monitoring”
• Updated timestamp: Shows when data was last refreshed (e.g., “Updated 16:24:46”)
• Refresh all data button: Click to manually refresh all dashboard metrics

---

​

The 4 Key Metrics at a Glance

​

1. System Health: 100%

Visual indicator: Green left border indicates healthy status When to worry: Below 95% means you should investigate Active Insights.

---

​

2. Active Insights: 3

What it shows: Number of issues currently requiring your attention Visual indicators: - Large number “3” shows total insights

• Badge “4” indicates newer/updated insights
• Orange icon for warnings

From our real dashboard: - “3” issues requiring attention

• Includes: ImagePullBackOff, CrashLoopBackOff, OOMKilled

Types of insights: - ** Types of insights:** - OOMKilled, critical failures

• medium severity - Pod crashes, image pull errors
• low severity - Configuration warnings

Click the card to go directly to Insights page for full details.

---

​

3. Intelligent Analysis: 3 RCA Reports

What it shows: Number of Root Cause Analysis reports generated by the RCA Pipeline Agent Visual indicators: - Number “3” shows completed RCA reports

• Shield icon indicates intelligent analysis
• Orange left border for attention

From our dashboard: - 3 RCA reports completed

• Shows analysis is done, not just detection

What this means: - RubixKube didn’t just detect failures

• It ANALYZED them and found root causes
• Complete reports available with evidence

Click to view all RCA reports in Insights.

---

​

4. Agents: 3/4 Active

What it shows: How many AI agents are running / total agents From our dashboard: “3/4 Active”

• 3 agents operational
• 1 agent offline or degraded

The 4 agents: 1.** RubixKube Observer**- Monitors cluster 2.RCA Pipeline - Analyzes incidents 3.Memory Agent - Stores learnings 4.SRI Agent - Powers Chat Visual indicator: Green gear icon shows agents are mostly healthy When not all active: Click Agents in navigation to see which one is offline.

---

​

Infrastructure Panel

Status indicator: Connected (healthy cluster connection) Panel controls: - ** Panel controls:** button (circular arrow icon) - Update infrastructure numbers

• Show details button - Expand to see additional metrics
• View page link - Go to full Infrastructure topology

---

​

Infrastructure Metrics Overview

​

Summary metrics:

Visual indicators: - Server icon for Nodes

• Cube icon for Pods
• Grid icon for Deployments
• Globe icon for Services

---

​

Expanded Infrastructure View

​

Click “Show details” to reveal:

When expanded: - Button changes to “Hide details”

• Two additional metrics appear at the bottom
• All resource counts remain visible

Use case: Get a complete infrastructure overview without leaving the dashboard

---

​

Activity Feed: Real-Time Event Stream

​

From our real dashboard - 7 events showing:

​

Event Types You’ll See

---

​

Understanding Event Details

Each event card shows: Top row: - ** Top row:** - Indicates event type (warning triangle, magnifying glass, etc.)

• Title - Clear description of what happened
• Badge - INSIGHT or RCA tag

Second row: - ** Second row:** - “Severity: high/medium/low” or “RCA Report Analysis Complete” Bottom row: - ** Bottom row:** - When the event occurred (MM/DD/YYYY HH:MM AM/PM)

• Status badge - “active” or “resolved”

Color coding: - Blue left border = INSIGHT events

• Purple left border = RCA events
• Red “high” text = Critical severity

The feed updates in real-time - no refresh needed!

---

​

RCA Live Stream

Current state from our dashboard: - Status:** Current state from our dashboard:** - Status:** Current state from our dashboard:** - Message: “No recent RCA events”

• Instruction: “Connect to SSE to see live events”

When active, you’ll see: - Function calls being executed

• Analysis percentage (e.g., “60% complete”)
• Evidence being gathered
• Pattern matching in progress
• Real-time streaming of RCA analysis steps

Statuses: - ** Statuses:** - Connected to SSE stream, ready for analysis

• Offline - Not connected to event stream
• Shows event count when active analysis is running

Use case: Watch RCA Pipeline Agent work in real-time as it investigates incidents

---

​

Dashboard Workflow: Your Daily Routine

​

Morning Check (60 seconds)

Result: Complete understanding of cluster state in under 1 minute.

---

​

Real Examples from Our Dashboard

​

Example 1: OOMKilled Detected (HIGH Severity)

​

What appeared in Activity Feed:

“Out of memory (OOMKilled) detected on a pod in Pod/memory-hog-demo”

• Type: INSIGHT
• Severity : high(red text)
• Status: active
• Time: 01:58 AM

What to do: 1. Note it’s HIGH severity (needs immediate attention) 2. Look for corresponding RCA report (should appear shortly after) 3. Click on the event for full details 4. Goes to Insights page with complete analysis 5. Review root cause and recommended fixes 6. Apply memory limits or increase resources Expected RCA: Should see a follow-up event like:

“Postmortem: OOMKilled Event on Pod ‘memory-hog-demo‘“

---

​

Example 2: CrashLoopBackOff with RCA

​

What appeared:

1.First - INSIGHT event: > “Container experiencing repeated crashes in crash-loop-demo (restart count: 3)”

• Type: INSIGHT
• Severity: medium
• Status: active

2.Then - RCA event: > “Postmortem: Pod ‘crash-loop-demo’ in CrashLoopBackOff State”

• Type: RCA
• Status: RCA Report Analysis Complete

What this pattern means: - RubixKube detected the crash loop (INSIGHT)

• RCA Pipeline automatically analyzed it (RCA)
• Complete report now available with root cause

Action: Click the RCA event to read the full analysis with evidence and recommendations.

---

​

Example 3: ImagePullBackOff

​

What appeared:

“Pod broken-image-demo has been pending for an extended period”

• Type: INSIGHT
• Severity: medium
• Status: active

What this means: - Pod stuck in Pending state

• Likely ImagePullBackOff error
• Not critical but needs fixing

Troubleshooting: - Check image name and tag

• Verify registry credentials
• Confirm image exists in registry

---

​

Understanding the Metrics

​

System Health Calculation

​

How RubixKube calculates the 100%:

Why 100% with 3 active insights? - The failing pods are in tutorial namespace (non-critical)

• Core system pods are all healthy
• System infrastructure is operational
• Agents are mostly active (3/4)

---

​

Infrastructure Panel Deep Dive

​

What each metric means:

​

Nodes: 1

Our cluster: Single-node KIND cluster In production: Would show multiple nodes (e.g., “5 nodes”) What to watch: - Node count changes (scale up/down)

• All nodes should be Ready status
• Node resource pressure (memory, CPU, disk)

Click “View page” to see:

• Node details
• Resource usage per node
• Taints and tolerations
• Node conditions

---

​

Pods: 14

Our cluster breakdown: - 8 system pods (kube-system namespace)

• 3 RubixKube pods (rubixkube-system)
• 3 failing tutorial pods (tutorial namespace)

Color indicators: - Green number = All healthy

• Yellow = Some issues (warning)
• Red = Critical failures

What to watch: - Sudden drops (pods crashing)

• Unusual increases (unexpected deployments)
• Pods stuck in Pending

Click metric to filter Infrastructure view to pods only

---

​

Deployments: 4

Our cluster: - coredns (DNS service)

• local-path-provisioner (storage)
• kubernetes-mcp-server (API)
• rubixkube-observer (monitoring agent)

What deployments show: - Desired vs actual replica counts

• Rollout status
• Pod template health

Click “View page” to see:

• Deployment health
• Replica counts (desired/current/ready)
• Rollout history
• Pod template specs

---

​

Services: 3

Our cluster: - kubernetes (API server)

• kube-dns (DNS)
• rubixkube-observer (monitoring)

Service types: - ** Service types:** - Internal cluster access

• NodePort - External access via node IP
• LoadBalancer - Cloud load balancer

What to watch: - Service endpoints (should have backing pods)

• Port configurations
• Service discovery issues

---

​

Namespaces: 7 (Expanded View)

​

Click “Show details” to see this metric

Our cluster namespaces: 1. default 2. kube-system (system pods) 3. kube-public 4. kube-node-lease 5. rubixkube-system (RubixKube components) 6. tutorial (demo failing pods) 7. local-path-storage Why this matters: - Understand cluster organization

• Identify namespace-specific issues
• Track namespace growth

---

​

Agents: 1 (Expanded View)

​

Click “Show details” to see this metric

Shows: Number of RubixKube agent pods deployed Our cluster: 1 agent pod (rubixkube-observer) Cross-reference: Compare with “Agents: 3/4” metric

• ** 3/4 Active** = 3 of 4 agent types are running
• ** 1 Agent pod** = 1 physical pod deployed (may run multiple services)

---

​

Activity Feed Explained

​

From our dashboard - 7 events in chronological order (newest first):

1.Pod broken-image-demo pending (INSIGHT, medium, 01:59 AM) 2.crash-loop-demo crashes (INSIGHT, medium, 01:58 AM) 3.crash-loop-demo RCA complete (RCA, 01:58 AM) 4.memory-hog-demo crashes (INSIGHT, medium, 01:58 AM) 5.memory-hog-demo RCA complete (RCA, 01:58 AM) 6.OOMKilled on memory-hog-demo (INSIGHT, HIGH, 01:58 AM) ⭐ Most critical 7.OOMKilled RCA complete (RCA, 01:58 AM) Notice the pattern: - INSIGHT first (detection by Observer)

• RCA follows within seconds (analysis by RCA Pipeline)
• Chronological order helps track incident progression

Feed features: - ** Feed features:** - Events appear as they occur

• Persistent scrollback - Historical events remain
• Badge indicators - Easy visual scanning
• Click to expand - View full event details

---

​

Dashboard Actions You Can Take

​

1. Refresh All Data

Button: Top right “Refresh all data” (circular arrow icon) What it does: - Re-queries all metrics

• Updates infrastructure counts
• Refreshes activity feed
• Updates agent status
• Reloads RCA stream connection

When to use: - Manually verify latest state

• After making cluster changes
• When troubleshooting stale data

Visual feedback: Button briefly shows active state, timestamp updates

---

​

2. Refresh Infrastructure Only

Button: Infrastructure panel “Refresh data” button What it does: - Updates only Infrastructure panel metrics

• Faster than full dashboard refresh
• Doesn’t reload Activity Feed or metrics

When to use: - After deploying new resources

• Scaling deployments
• Adding/removing nodes

---

​

3. Expand Infrastructure Details

Button: “Show details” / “Hide details” What it reveals: - Namespaces count

• Agents count
• Keeps existing metrics visible

When to use: - Need complete infrastructure overview

• Investigating namespace issues
• Verifying agent deployment

---

​

4. Navigate to Details

From metrics cards: - ** From metrics cards:** → Go to Insights page

• Click “Agents” card → Go to Agents page
• Number indicators are clickable

From Infrastructure: - ** From Infrastructure:** → Full Infrastructure topology

• Click individual metrics → Filtered views (when on Infrastructure page)

From Activity Feed: - ** From Activity Feed:** → Full event details and related insights

• Click INSIGHT badge → View insight details
• Click RCA badge → View RCA report

---

​

Common Dashboard Scenarios

​

Scenario 1: System Health Drops to 92%

What you see: - Health metric turns yellow/orange

• Active Insights count increased
• New events in Activity Feed

What to do:

---

​

Scenario 2: Active Insights Jumps from 0 to 3

What you see:

• Insights card shows “3” with badge “4”
• Multiple new events in Activity Feed
• Timestamps are recent (last few minutes)

What to do:

---

​

Scenario 3: Agent Goes Offline (shows 2/3)

What you see:

• Agents metric shows “2/3” or “3/4”
• May see “Agent health degraded” in Activity Feed
• System Health may decrease slightly

What to do:

---

​

Scenario 4: RCA Live Stream Shows Activity

What you see:

• RCA Live Stream panel shows “LIVE” with event count (e.g., “3 events”)
• Events stream in real-time
• May see function calls, analysis steps

What to do:

---

​

Scenario 5: Infrastructure Metrics Change Unexpectedly

What you see:

• Pod count increased from 14 to 18
• No corresponding Activity Feed events
• System Health still 100%

What to do:

---

​

Best Practices for Dashboard Usage

1. Start Every Day Here

Morning routine:

1

Open Dashboard

Make it your first tab every morning

2

60-Second Scan

Check metrics, scan feed, verify health

3

Note Changes

Compare to yesterday’s baseline

4

Drill Down

Investigate any anomalies immediately

Benefit: Early detection before customers notice issuesPro tip: Bookmark the dashboard URL for instant access

2. Keep Dashboard Open

Pro tip: Keep dashboard in a pinned browser tabWhy? - Real-time updates via Server-Sent Events (SSE)

• Activity Feed refreshes automatically
• Spot new incidents immediately as they occur
• No manual refresh needed

Setup: 1. Open Dashboard in browser 2. Right-click tab → “Pin Tab” 3. Position as first tab 4. Check throughout the dayBonus: Browser can show notification badge for new high-severity events

3. Learn Your Baseline

​

Track normal patterns over time:

System Health: - ** Normal range**: 95-100%

• Typical: 100% for healthy clusters
• Alert threshold: Below 95%

Active Insights: - ** Normal range**: 0-2 for healthy clusters

• Alert threshold: 3+ or sudden increase
• Critical: 5+ insights

Infrastructure: - ** Your normal pod count**: ____

• Your normal namespace count: ____
• Expected agent status: 3/3 or 4/4

Benefit: Anomalies stand out immediately when you know your baselineAction: Take note of current values, revisit weekly

4. Use Activity Feed as Audit Log

​

The feed is chronological and persistent:

Use cases: - See when issues started (timestamp tracking)

• Track incident resolution timeline
• Verify deployments and changes
• Post-mortem analysis
• Compliance and reporting

Features: - Events remain in feed (not just transient)

• Clickable for full details
• Searchable (future feature)
• Exportable for reports

Pro tip: Take screenshots of Activity Feed during incidents for documentation

5. Understand the Insight → RCA Pattern

​

Learn to recognize the two-step pattern:

Step 1: INSIGHT appears - Blue badge, INSIGHT tag

• RubixKube Observer detected an issue
• Severity assigned (high/medium/low)
• Status: active

Step 2: RCA follows (30-60 seconds later) - Purple badge, RCA tag

• RCA Pipeline analyzed the issue
• Root cause identified
• Status: RCA Report Analysis Complete

What this means: - Don’t panic when you see an INSIGHT

• Wait for the RCA before taking action
• RCA provides context and recommendations
• Pattern shows system is working correctly

Action: Train your team to look for both events

6. Triage by Severity

​

Priority order for responding to insights:

** 1. HIGH severity (red text)** - OOMKilled

• Critical pod failures
• Security issues
• Data loss risk
• Action: Immediate response required

** 2. medium severity** - CrashLoopBackOff

• ImagePullBackOff
• Configuration errors
• Action: Investigate within hours

** 3. low severity** - Warnings

• Deprecations
• Minor issues
• Action: Plan fix in next sprint

Dashboard makes this easy: - HIGH severity shows in red

• Easy visual scanning
• Click to investigate details

---

​

What You Learned

---

​

Quick Reference

​

Dashboard at a glance:

---

​

Real Data from Our Dashboard

What we captured: - System Health:** What we captured:** (all core systems healthy)

• Active Insights:** 3 ** (broken-image-demo, crash-loop-demo, memory-hog-demo)
• Intelligent Analysis:** 3 RCA reports** (all analyzed with root causes)
• Agents:** 3/4 active** (1 offline, 3 operational)
• Infrastructure:** 1 node, 14 pods, 4 deployments, 3 services ** - Expanded:** 1 node, 14 pods, 4 deployments, 3 services ** - Expanded:** 1 node, 14 pods, 4 deployments, 3 services** (4 INSIGHTS, 3 RCA reports)
• RCA Stream:LIVE with 0 active events

This demonstrates: - Dashboard provides complete visibility even when things are failing

• System distinguishes between critical vs non-critical failures
• RCA Pipeline automatically analyzes incidents
• Real-time monitoring with historical context

---

​

Interactive Elements Summary

​

Clickable actions:

---

​

Keyboard Shortcuts

While on Dashboard:

• R - Refresh all data
• I - Navigate to Insights
• A - Navigate to Agents
• Esc - Close any open modals

(Note: Keyboard shortcuts may vary by implementation)

---

​

Troubleshooting Dashboard Issues

​

Dashboard not loading

Symptoms: Spinner stays visible, “Initializing dashboard…” message persists Causes: - Backend API connection issue

• SSE connection failed
• Browser cache problem

Solutions: 1. Hard refresh: Cmd+Shift+R (Mac) or Ctrl+Shift+R (Windows) 2. Check browser console for errors 3. Verify cluster connection in Settings 4. Check RubixKube backend pods are running

---

​

Activity Feed shows “No recent activity” but issues exist

Symptoms: Feed is empty despite active insights Causes: - SSE connection not established

• Events not being generated
• Time range filter active

Solutions: 1. Click “Refresh all data” 2. Check browser console for SSE errors 3. Verify RubixKube Observer is running 4. Check “Connect to SSE to see live events” message

---

​

Metrics showing old data

Symptoms: Timestamp not updating, stale numbers Causes: - Auto-refresh disabled

• API timeout
• Backend issue

Solutions: 1. Click “Refresh all data” manually 2. Check network tab for failed requests 3. Verify backend API health 4. Restart browser if needed

---

​

Infrastructure shows “Disconnected”

Symptoms: Red status indicator, no metrics Causes: - Cluster connection lost

• Kubeconfig invalid
• Network issue

Solutions: 1. Go to Settings → Clusters 2. Verify cluster connection status 3. Test connection 4. Re-add cluster if needed 5. Check kubeconfig validity

---

​

Performance Tips

For large clusters: - Dashboard optimized for up to 1000 pods

• Activity Feed shows most recent 50 events
• Older events paginated
• Infrastructure panel uses aggregated counts

For slow connections: - Dashboard data compressed

• SSE uses efficient streaming
• Images lazy-loaded
• Metrics cached briefly

For multiple clusters: - Switch cluster in Settings

• Dashboard refreshes automatically
• Each cluster has independent metrics
• Activity Feed filtered by selected cluster

---

​

Next Steps

---

​

You’re now a Dashboard expert! Use it daily to stay ahead of incidents.

---

​

Related Documentation

---

​

Need Help?

---

​

Feedback

Found an issue with this guide or have suggestions? We’d love to hear from you!

• Email: [email protected]
• Subject: “Dashboard Guide Feedback”

---

Last updated: October 6, 2025 Guide version: 2.0 Based on RubixKube Console v1.0

---

​

Related Guides

• Insights
• Infrastructure
• Agents
• Clusters
• Workspace
• Settings
• Integrations