Page 6 of 15

Following a Request Across Services

See how a single request_id tracks the entire journey from frontend to backend.

MDC Propagation Flow

The Power of Correlation IDs

One ID to Rule Them All

A correlation ID (we call it request_id) is a unique identifier that follows a request through its entire journey across all services.

This simple concept is the foundation of distributed tracing in microservices.

How request_id Flows Through Our System

Step 1: Browser → Frontend

  • User makes request to Frontend (e.g., create a person)
  • RequestLoggingFilter intercepts the request
  • Generates UUID: abc-def-123-456
  • Sets MDC values: request_id, user_ip, http_method, request_uri
  • All Frontend logs now include these values automatically

Step 2: Frontend → Backend

  • Frontend needs to call Backend service
  • MdcPropagationInterceptor intercepts RestTemplate call
  • Reads MDC values from current thread
  • Adds HTTP headers:
    • X-Request-ID: abc-def-123-456
    • X-Person-User-IP: 192.168.1.10
    • X-Person-HTTP-Method: POST
    • X-Person-Request-URI: /people
  • Sends HTTP request with these headers

Step 3: Backend Receives Request

  • Backend receives HTTP request with headers
  • RequestLoggingFilter intercepts the request
  • Checks for X-Request-ID header
  • If present: uses propagated value abc-def-123-456
  • If absent: generates new UUID
  • Sets MDC with same values
  • All Backend logs now include the same request_id!

Step 4: Result - Unified Logging

  • Frontend logs: request_id: abc-def-123-456
  • Backend logs: request_id: abc-def-123-456
  • Same ID in logs from BOTH services!
  • Query Kibana for this ID → see complete request journey

Querying in Kibana

Find All Logs for One Request
request_id: "abc-def-123-456"

This single query returns:

  • Frontend logs: request received, validation, calling backend
  • Backend logs: request received, database query, response sent
  • Complete timeline of events
  • All context: user IP, HTTP method, endpoint

Result: See the entire request journey in chronological order!

Real-World Example

Debugging a Failed Request

Scenario: User reports "I tried to create a person but got an error"

Without request_id:

  • Search Frontend logs for user's IP
  • Find timestamp of error
  • SSH into 3 Backend instances
  • Grep logs around that timestamp
  • Try to correlate manually
  • Time: 30+ minutes, lots of guessing

With request_id:

  • Copy request_id from error response
  • Query Kibana: request_id: "abc-def-123-456"
  • See complete flow: Frontend → Backend → Database
  • Find root cause: Database constraint violation
  • Time: 2 minutes, no guessing

Implementation Highlights

RequestLoggingFilter
  • Lives in spring-boot-common module
  • Shared by Frontend and Backend
  • Checks for propagated headers first
  • Generates new UUID if not present
  • Sets MDC values automatically
MdcPropagationInterceptor
  • Lives in spring-boot-front module
  • Implements ClientHttpRequestInterceptor
  • Intercepts all RestTemplate calls
  • Reads MDC, adds as HTTP headers
  • Transparent to application code

Key Takeaways

  • Single request_id tracks entire request journey across services
  • HTTP headers propagate context between Frontend and Backend
  • Query Kibana for request_id to see complete flow
  • Distributed tracing made simple with MDC + HTTP headers
  • Debugging time reduced from 30+ minutes to under 2 minutes