# Request ID Logging
## Overview
Every HTTP request to Mimir now gets a unique request ID that's tracked throughout the entire request lifecycle. This makes debugging and tracing much easier.
## What Was Added
### 1. Request ID Middleware (`mimir/middleware.py`)
- Generates a unique UUID for each request
- Can accept `X-Request-ID` header from upstream systems (load balancers, API gateways)
- Stores request ID on the request object as `request.request_id`
- Returns request ID in response headers as `X-Request-ID`
- Logs request start/end/errors with the request ID
### 2. Logging Filter (`mimir/logging_filters.py`)
- Adds request ID to every log message
- Shows `NO_REQUEST` for logs outside of HTTP requests (management commands, background tasks)
### 3. Thread-Local Storage (`mimir/request_local.py`)
- Stores current request in thread-local storage
- Allows logging filter to access request ID from anywhere in the codebase
## Log Format
All logs now include request ID:
**File logs (verbose):**
```
[2025-11-28 10:30:45] [INFO ] [REQ:a1b2c3d4-5678-90ef-ghij-klmnopqrstuv] [PID:12345:TID:67890] [accounts/views.py:85] [views.login_view] User maria attempting login
```
**Console logs (simple):**
```
[REQ:a1b2c3d4-5678-90ef-ghij-klmnopqrstuv] INFO User maria logged in successfully
```
## Usage
### In Views
Access the request ID directly from the request object:
```python
def my_view(request):
request_id = request.request_id
logger.info(f"Processing request for user {request.user.username}")
# The log message will automatically include the request ID
```
### For Client-Side Tracking
The response includes the request ID in headers:
```bash
curl -I http://localhost:8000/dashboard/
# Response includes:
# X-Request-ID: a1b2c3d4-5678-90ef-ghij-klmnopqrstuv
```
### Propagating from Upstream
If you have a load balancer or API gateway generating request IDs, pass it along:
```bash
curl -H "X-Request-ID: my-custom-tracking-id" http://localhost:8000/api/
# Response will echo back the same ID:
# X-Request-ID: my-custom-tracking-id
```
### Searching Logs
To trace a specific request through all logs:
```bash
# Find all log entries for a specific request
grep "REQ:a1b2c3d4-5678-90ef-ghij-klmnopqrstuv" logs/app.log
# See the full journey of a request
grep "REQ:a1b2c3d4" logs/app.log | grep -E "(REQUEST_START|REQUEST_END|REQUEST_ERROR)"
```
## Benefits
1. **Easy Debugging**: Follow a single request through multiple log statements
2. **Error Tracking**: When users report errors, ask for the request ID from browser dev tools
3. **Performance Analysis**: Track how long specific requests take
4. **Distributed Tracing**: Compatible with upstream tracing systems
5. **User Support**: Users can provide request ID when reporting issues
## Example Request Flow
```
1. User hits /auth/user/login/
[REQ:abc-123] [REQUEST_START] GET /auth/user/login/ - User: anonymous
2. Login form displayed
[REQ:abc-123] Displaying login form
3. User submits credentials
[REQ:def-456] [REQUEST_START] POST /auth/user/login/ - User: anonymous
[REQ:def-456] Login attempt for username: maria, remember_me: True
[REQ:def-456] User maria logged in with remember me (30 days session)
4. Redirect to dashboard
[REQ:def-456] [REQUEST_END] POST /auth/user/login/ - Status: 302
5. Dashboard loads
[REQ:ghi-789] [REQUEST_START] GET /dashboard/ - User: maria
[REQ:ghi-789] User maria accessing dashboard
[REQ:ghi-789] [REQUEST_END] GET /dashboard/ - Status: 200
```
## Configuration
All configuration is in `mimir/settings.py`:
```python
# Middleware order matters - RequestIDMiddleware should be after auth
MIDDLEWARE = [
...
"mimir.middleware.RequestIDMiddleware",
]
# Logging config includes request_id filter
LOGGING = {
'filters': {
'request_id': {
'()': 'mimir.logging_filters.RequestIDFilter',
},
},
'handlers': {
'file': {
'filters': ['request_id'],
...
},
},
}
```
## Testing
Run the request ID integration tests:
```bash
pytest tests/integration/test_request_id_logging.py -v
```
Tests verify:
- Request IDs are generated and returned in headers
- Custom request IDs are propagated
- Each request gets a unique ID
- Request IDs are accessible in views
## Performance Impact
Minimal - adds approximately:
- UUID generation: ~1-2μs per request
- Thread-local storage access: ~0.1μs per log statement
- Header addition: ~0.1μs per request
Total overhead: < 5μs per request (negligible).
