Django Patchy

Useful django app and utils to develop large-scale web app

Requirements

• Python (2.7)
• Django (1.8, 1.9)

Installation

Install using pip...

pip install djangopatchy

Middlewares

LongRequestMiddleware

Let's add a middleware to calculate the time elapse for each request

1. Add a middleware at the top:

MIDDLEWARE_CLASSES = (
    'patchy.middleware.LongRequestMiddleware',
    ...
)

2. Add logger handler with pathcy.middleware:

Example:

'patchy.middleware': {
    'handlers': ['sentry'],
    'level': 'ERROR',
    'propagate': True
}

3. Set the timeout threshold in settings(default to 1 second):

PATCHY_LONG_REQUEST_TIMEOUT = 2  # set the timeout to 2 seconds

Results:

• For each web request, it has a header variable X-ELAPSED in seconds to indicate the time elapse.

X-ELAPSED: 0.005 # it means the request costs 5 ms

• If it exceeds the PATCHY_LONG_REQUEST_TIMEOUT a error log message will be sent.

4. Add some urls to ignore list(optional), regular expression is supported.

PATCHY_LONG_REQUEST_IGNORE_URLS = [
    r'^/_admin/.*$',
]

Utilities

long_sql_execute_wrapper

Let us rewrite the CursorWrapper.execute to calculate the sql process time

1. Add the python snippets in the djangoproject/__init__.py

# rewrite the sql operation method
from django.db.backends import utils
from patchy.utils import long_sql_execute_wrapper
utils.CursorWrapper.execute = long_sql_execute_wrapper

2. Add logger handler with pathcy.utils:

Example:

'patchy.utils': {
    'handlers': ['sentry'],
    'level': 'ERROR',
    'propagate': True
}

3. Set the timeout threshold in settings(default to 0.05 seond, which is 50 miliseconds):

PATCHY_LONG_SQL_TIMEOUT = 0.01  # set the timeout to 10 miliseconds

Result:

• If the sql operation exceeds the PATCHY_LONG_SQL_TIMEOUT a error log message will be sent.

4. no_sql_monitoring

no_sql_monitoring is a decorator for wrapping code to skip sql monitoring.

from patchy.utils import no_sql_monitoring

@no_sql_monitoring
def transaction_status(request, transaction_no):
    transaction = Transaction.objects.get(transaction_no=transaction_no)

    return ResponseBuilder.build_success_json_response(
        msg='查询订单成功',
        data={
            'transaction_no': transaction_no,
            'transaction_status': transaction.transaction_status
    })