Skip to main content
The Predictive Maintenance system uses InfluxDB Cloud as its primary time-series database for storing sensor data, features, anomalies, and operator logs.

​
Why InfluxDB?

InfluxDB is optimized for time-series data with:
  • High write throughput for sensor ingestion (100Hz+ supported)
  • Efficient storage with automatic downsampling and retention policies
  • Powerful querying using Flux language for feature engineering
  • Cloud-managed infrastructure (no server maintenance)
  • Free tier with 30-day retention and 10,000 write requests/day

​
Setup Process

1

Create InfluxDB Cloud account

Sign up for a free account at InfluxData Cloud.Recommended region: AWS us-east-1 (lowest latency for US-based deployments)
2

Create a bucket

Navigate to Load Data > Buckets in the InfluxDB UI.Click + Create Bucket and configure:
  • Name: sensor_data (must match backend configuration)
  • Retention: 30 days (free tier) or custom
  • Description: Time-series data for predictive maintenance
The bucket name sensor_data is hardcoded in the default configuration. If you use a different name, update the INFLUX_BUCKET environment variable.
3

Generate an API token

Go to Load Data > API Tokens and click + Generate API Token.Select All Access Token to grant read/write permissions.
Copy the token immediately! It will only be displayed once. If you lose it, you’ll need to generate a new token.
Store it securely in a password manager or environment file.
4

Get your organization ID

Click your user profile in the top-right corner and select About.Copy the Organization ID (a hexadecimal string like a1b2c3d4e5f6g7h8).
5

Find your instance URL

The URL is shown in the browser address bar:
https://us-east-1-1.aws.cloud2.influxdata.com
Common regions:
  • AWS US East: https://us-east-1-1.aws.cloud2.influxdata.com
  • AWS US West: https://us-west-2-1.aws.cloud2.influxdata.com
  • AWS EU Central: https://eu-central-1-1.aws.cloud2.influxdata.com

​
Environment Configuration

​
Local Development

Create a backend/.env file (or .env in the project root): 
# InfluxDB Cloud Configuration
INFLUX_URL=https://us-east-1-1.aws.cloud2.influxdata.com
INFLUX_TOKEN=your-api-token-here
INFLUX_ORG=a1b2c3d4e5f6g7h8
INFLUX_BUCKET=sensor_data

# Optional: Application Settings
ENVIRONMENT=local
Never commit .env files to version control! Add .env to your .gitignore to prevent accidental exposure of credentials.

​
Docker Compose

Add environment variables to docker-compose.yml: 
services:
  backend:
    build: .
    environment:
      - INFLUX_URL=https://us-east-1-1.aws.cloud2.influxdata.com
      - INFLUX_TOKEN=${INFLUX_TOKEN}
      - INFLUX_ORG=${INFLUX_ORG}
      - INFLUX_BUCKET=sensor_data
    env_file:
      - .env
Then create a .env file in the same directory: 
INFLUX_TOKEN=your-api-token-here
INFLUX_ORG=a1b2c3d4e5f6g7h8

​
Production Deployment (Render)

In the Render dashboard, go to your service settings and add environment variables:
KeyValue
INFLUX_URLhttps://us-east-1-1.aws.cloud2.influxdata.com
INFLUX_TOKENyour-api-token-here
INFLUX_ORGa1b2c3d4e5f6g7h8
INFLUX_BUCKETsensor_data
ENVIRONMENTproduction
Render automatically redeploys your service when environment variables are updated.

​
Vercel Frontend (Optional)

The frontend doesn’t need direct InfluxDB access (it queries the backend API), but you can configure the API URL: 
VITE_API_URL=https://predictive-maintenance-uhlb.onrender.com

​
Verification

​
Test Database Connection

Use the FastAPI health endpoint to verify InfluxDB connectivity: 
curl http://localhost:8000/health
Expected response: 
{
  "status": "healthy",
  "db_connected": true
}
If db_connected: false, check:
  1. Token validity: Regenerate if expired
  2. Organization ID: Verify it matches your account
  3. Bucket existence: Ensure sensor_data bucket is created
  4. Network access: Check firewall rules (InfluxDB Cloud uses HTTPS/443)

​
Manual Query Test

You can test the connection directly using the InfluxDB Python client: 
from influxdb_client import InfluxDBClient
import os

url = os.getenv("INFLUX_URL")
token = os.getenv("INFLUX_TOKEN")
org = os.getenv("INFLUX_ORG")
bucket = os.getenv("INFLUX_BUCKET")

client = InfluxDBClient(url=url, token=token, org=org)

# Test query
query_api = client.query_api()
query = f'from(bucket: "{bucket}") |> range(start: -1h) |> limit(n: 5)'
result = query_api.query(query)

for table in result:
    for record in table.records:
        print(record.values)

client.close()

​
Data Schema

The system writes data to InfluxDB using these measurements:

​
sensor_data (Raw Sensor Readings)

FieldTypeUnitDescription
voltage_vfloatVoltsLine voltage
current_afloatAmperesLine current
power_factorfloatUnitlessPower factor (0-1)
vibration_gfloatg-forceAcceleration
power_kwfloatKilowattsActive power
Tags:
  • asset_id: Unique asset identifier (e.g., motor_01)
  • operating_state: RUNNING, IDLE, or OFF

​
features (Engineered Features)

FieldDescription
voltage_rolling_mean_1h1-hour moving average of voltage
current_spike_countNumber of current spikes in 10-point window
power_factor_efficiency_scoreNormalized efficiency (0-100)
vibration_intensity_rmsRMS vibration intensity
voltage_stabilityDeviation from 230V nominal
power_vibration_ratioVibration per unit power factor

​
anomalies (ML Predictions)

FieldDescription
anomaly_scoreIsolation Forest score (0-1)
batch_anomaly_scoreBatch model score (0-1)
health_scoreOverall health (0-100)
risk_levelLOW, MODERATE, HIGH, CRITICAL
degradation_indexCumulative damage (0-1)
damage_rateRate of health decline (per hour)
rul_hoursRemaining useful life (hours)

​
operator_logs (Maintenance Events)

FieldDescription
messageEvent description
severityINFO, WARNING, CRITICAL
technicianOperator name

​
Retention and Downsampling

​
Free Tier Limits

  • Retention: 30 days
  • Write rate: 10,000 writes/day
  • Query rate: Unlimited
  • Storage: 10 MB
At 1 write/second (1Hz), you’ll hit the free tier limit in ~2.8 hours/day. For 24/7 operation, upgrade to the Usage-Based Plan ($0.002 per 1,000 writes).

​
Downsampling Task (Optional)

To reduce storage, create a downsampling task that aggregates 1-second data into 1-minute averages: 
option task = {name: "Downsample Sensor Data", every: 1h}

from(bucket: "sensor_data")
  |> range(start: -1h)
  |> filter(fn: (r) => r._measurement == "sensor_data")
  |> aggregateWindow(every: 1m, fn: mean, createEmpty: false)
  |> to(bucket: "sensor_data_downsampled", org: "your-org")
This reduces write volume by 60x while preserving trends.

​
Troubleshooting

​
Error: “unauthorized access”

Cause: Invalid or expired API token. Solution:
  1. Regenerate a new API token in the InfluxDB UI
  2. Update the INFLUX_TOKEN environment variable
  3. Restart the backend service

​
Error: “bucket not found”

Cause: The INFLUX_BUCKET variable doesn’t match the actual bucket name. Solution:
  • Verify bucket name in InfluxDB UI matches sensor_data
  • Or update INFLUX_BUCKET to match your custom bucket name

​
Error: “organization not found”

Cause: Incorrect INFLUX_ORG value. Solution:
  1. Go to InfluxDB UI > Profile > About
  2. Copy the Organization ID (not the organization name)
  3. Update INFLUX_ORG with the correct ID

​
High Write Latency

Cause: Cross-region latency (e.g., EU backend writing to US-East InfluxDB). Solution:
  • Use an InfluxDB region close to your backend deployment
  • Enable batching in the write API (automatically handled by the backend)

​
Security Best Practices

Production Security Checklist

  • ✅ Use All Access Tokens only in secure environments
  • ✅ Generate scoped tokens (read-only, write-only) for specific services
  • ✅ Rotate tokens every 90 days
  • ✅ Store tokens in secret managers (AWS Secrets Manager, Vercel Secrets)
  • ✅ Never log token values in application code
  • ✅ Use environment variables, not hardcoded strings
  • ✅ Enable IP allowlisting in InfluxDB Cloud (Enterprise plan)

​
Next Steps

Calibration Workflow

Configure baseline models using stored data

Data Generator

Populate InfluxDB with test sensor data

Build docs developers (and LLMs) love

Get started for freeTalk to us