🖥️ UserAcquisition Infrastructure Dashboard

✅ All Systems Operational = Everything is fine, no action needed
🔴 System Needs Attention = Something is wrong, check details and escalate

Requests (12h) — Total requests handled in last 12 hours
5xx Errors — Server errors. Should be 0. If > 0, escalate immediately
CPU % — How busy the servers are. Green < 50%, Yellow 50-80%, Red > 80%
Memory % — RAM usage. Green < 80%
Response Time — How fast the server responds. Green < 0.5s, Yellow 0.5-1s, Red > 1s
Running Tasks — Number of servers running (min 10, max 100)

Each row = 30 minutes of data

Requests — Total hits in that 30-min window
Req/Target — Requests per ALB target (auto-scaling triggers at 100)
Req/Task — Requests divided by number of tasks. Shows load per server
Tasks — How many servers were running
CPU % — Average CPU across all tasks. Scale-up triggers at 50%
Memory % — Average memory usage across all tasks
Response (s) — Average response time in seconds
5xx Errors — Server errors in that window. Should always be 0

Color coding: 🟢 Green = healthy, 🟡 Yellow = warning, 🔴 Red = critical

Shows each ECS container (server) currently running

Task ID — Unique ID (first 8 chars)
Private IP — Internal IP address of the container
Status — RUNNING (green) = healthy. Anything else = problem
Req/Task (5m) — How many requests this task handled in last 5 minutes
Uptime — How long this task has been running since last deployment

The system automatically adds/removes servers based on:

CPU > 50% → Adds more tasks (scale-out in 30 seconds)
Requests > 100 per target per minute → Adds more tasks
CPU drops → Removes extra tasks after 5 minutes

Min: 10 tasks | Max: 100 tasks | Current task: 1 vCPU, 2 GB RAM

Escalate to Kalpesh immediately if:

❌ Red status banner
❌ 5xx errors > 0 for more than 5 minutes
❌ CPU > 80% sustained
❌ Tasks < 10 (some went down)
❌ Unhealthy targets > 0
❌ Response time > 2 seconds sustained
❌ All tasks showing 0 requests (traffic stopped)

NOT an issue: Single 5xx error, brief CPU spike, tasks going to 11-12 temporarily (that's auto-scaling working)