We Keep Coding

Internal Server Error for Hello World django project in production

I have my apache server directing traffic from mysite.com/home to my wsgi.py file, but when get_wsgi_application() is called I get an internal server error.
Running apache2 on amazon linux EC2. In wsgi.py, I've replaced the line,
application = get_wsgi_application()
with the function,
def application(environ, start_response):
status = '200 OK'
output = b'sys.path = %s\n' % repr(sys.path).encode('UTF-8')
output += b'sys.version = %s\n' % repr(sys.version).encode('UTF-8')
output += b'sys.prefix = %s\n' % repr(sys.prefix).encode('UTF-8')
response_headers = [('Content-type', 'text/plain'),
('Content-Length', str(len(output)))]
start_response(status, response_headers)
return [output]
and the server error goes away. It shows that apache is using the correct python version (3.7) and doesn't return an error when I import get_wsgi_application() from django.core.wsgi. To my understanding this means apache is finding the access point to my django project just fine, and it has permission to look into the file and show its contents. What am I missing?
Other possibly relevant information I can think of:
user "apache" and group "apache" own the outer mysite directory and everything in it. I've set all the settings in settings.py for production according to the django deployment checklist. https://docs.djangoproject.com/en/2.1/howto/deployment/checklist/
wsgi.py:
import os, sys
from django.core.wsgi import get_wsgi_application
os.environ.setdefault('DJANGO_SETTINGS_MODULE', 'mysite.settings')
sys.path.append('/usr/local/wsgi/django')
sys.path.append('/usr/local/wsgi/django/mysite')
application = get_wsgi_application()
httpd.conf:
<VirtualHost *:80>
ServerName www.mysite.com
Redirect permanent / httpswww.mysite.com
</VirtualHost>
<VirtualHost *:443>
ServerName httpswww.mysite.com
ServerAlias mysite.com
ServerAdmin person#gmail.com
DocumentRoot "/var/www/html/efs/Website"
WSGIDaemonProcess mysite
WSGIProcessGroup mysite
WSGIScriptAlias /home /usr/local/wsgi/django/mysite/mysite/wsgi.py
<Directory "/usr/local/wsgi/django/mysite/mysite">
<Files "wsgi.py">
Require all granted
</Files>
</Directory>
SSLEngine on
SSLCertificateFile "/crt/path.crt"
SSLCertificateKeyFile "/key/path.key"
</VirtualHost>
settings.py:
SECRET_KEY = 'string_I_havent_moved_yet'
DEBUG= False
ALLOWED_HOSTS = ['www.mysite.com', e2-serial-numbers.compute-1.amazonaws.com']

Flask WSGI subdomain dispatching causing 404s on app instances

My goal is for my Flask app to switch databases based on the subdomain.
The subdomain dispatcher described in the docs is useful and I'm using something based off it
Everything seems to work fine but only 2 instances seem to work at a given time despite the SubDomainDispatcher having multiple instances loaded. The other instances give out 404s.
My create_app function is as follows:
def create_app(subdomain):
app = Flask("proj", static_folder="../static", template_folder="../templates")
with app.app_context():
app.config.from_object('proj.config')
app.config.update(
SQLALCHEMY_DATABASE_URI = 'mysql://root:password#localhost:3306/' + subdomain + '_proj',
HOST = subdomain + app.config["HOST"],
)
from proj.models import db, Models
engine = sqlalchemy.create_engine("mysql://root:passwordT#localhost:3306")
engine.execute("CREATE DATABASE IF NOT EXISTS "+subdomain+"_proj")
engine.execute("use "+subdomain+"_proj")
db.init_app(app)
db.create_all()
mail = Mail(app)
store = DictStore()
KVSessionExtension(store, app)
import proj.views
return app
I'm using Apache with mod_wsgi:
Apache Config
<VirtualHost *:80>
ServerName mywebsite.com
ServerAdmin admin#mywebsite.com
WSGIScriptAlias / /var/www/proj/app.wsgi
WSGIDaemonProcess application user=www-data group=www-data processes=5 threads=20
<Directory /var/www/proj/proj/>
Order allow,deny
Allow from all
</Directory>
#Alias /static /var/www/FlaskApp/FlaskApp/static
#<Directory /var/www/FlaskApp/FlaskApp/static/>
# Order allow,deny
# Allow from all
#</Directory>
ErrorLog ${APACHE_LOG_DIR}/error.log
LogLevel warn
CustomLog ${APACHE_LOG_DIR}/access.log combined
</VirtualHost>
wsgi file
import sys
sys.path.insert(0, '/var/www/proj')
from subdomain import SubdomainDispatcher
from proj import create_app
application = SubdomainDispatcher(create_app)
Seems like some sort of race condition but I don't see where or why it would be happening. I've tried reducing the amount of processes and threads to 1 each but that also did not see to work.
Any advice is greatly appreciated as I've been struggling with this for awhile.

After a few other eyes took a look, the problem is that imports will only run once causing routes to only ever be associated to the first app created.
Essentially views are only ever instantiated once for the first app/subdomain hit.
The solution is instead of importing you have function in each module initializing the routes.
As so in views.py:
def init_views(app):
#app.route('/page')
def page():
return "asdf"
Thus each app has it's own copy of the routes instead of only the first.
What a doozy.

Some routes in Flask Restful API not working

I've built an API in Flask, using Twilio's Restful extension. Everything works perfectly on Flask's development server. However once I move the app over to Apache and mod_wsgi, some routes stopped working
Apache config:
Listen 1337
<VirtualHost *:1337>
ServerName layer
ServerAdmin webmaster#localhost
DocumentRoot /var/www/layer
WSGIDaemonProcess layer user=this_is_me group=www-data threads=5
WSGIScriptAlias / /var/www/layer/app.wsgi
WSGIScriptReloading On
<Directory /var/www/layer>
WSGIProcessGroup layer
WSGIApplicationGroup %{GLOBAL}%
Options Indexes FollowSymLinks MultiViews
AllowOverride None
Order deny,allow
Allow from all
</Directory>
ErrorLog ${APACHE_LOG_DIR}/error.log
LogLevel info
CustomLog ${APACHE_LOG_DIR}/access.log combined
</VirtualHost>
In app.wsgi:
from src import app
application = app.create_app()
In app.py:
#!flask/bin/python
from flask import Flask, request, jsonify, Response
from flask.ext.restful import Resource, Api
from view import treeView, createBranchView, branchView, lineView, bulkView
def create_app():
app = Flask(__name__)
api = Api(app, catch_all_404s=True)
logging.basicConfig(filename=os.path.abspath('exlayer.log'), level=logging.DEBUG)
#Some stand alone routes
#app.route('/')
def index():
### Some code here ###
return jsonify({'status': 200, 'success':True})
#app.route('/create/', methods = ['POST'])
def create_tree():
### Some more code ###
return jsonify(dict(status=200, success=True))
## The rest of the routes
api.add_resource(treeView.dataTree, '/<tree_name>/')
api.add_resource(lineView.lineData, '/<tree_name>/line/<line_id>/')
api.add_resource(bulkView.bulkData, '/<tree_name>/bulk/<bulk_id>/')
api.add_resource(createBranchView.createBranch, '/<tree_name>/branch/')
api.add_resource(branchView.branchData, '/<tree_name>/branch/<branch_identifier>/')
app.config.from_pyfile('../config/config.py')
return app
In bulkView
So here's where things get interesting, if I send a get request to this route I get a Method Not Allowed 405 error. If I send a delete request it works fine. Virtually the same exact code runs in lineView without a problem.
from flask import Flask, request, jsonify, Response
from flask.ext.restful import Resource, Api
from src import helper
class bulkData(Resource):
def get(self, tree_name, bulk_id):
## Some code here ##
return jsonify({'status': 200, 'success':True})
def delete(self, tree_name, bulk_id):
## Some code here ##
return jsonify({'status': 200, 'success':True})
In branchView
Requests to either route concerning branches 404s. Checked the permissions on the files, tried splitting the classes up into separate files. No idea what's wrong :(

Use of os.path.abspath() and ../config/config.py will both cause problems because the working directory of your process will not be where your project is under Apache. Calculate absolute paths relative to os.path.dirname(file). See the mod_wsgi documentation:
http://code.google.com/p/modwsgi/wiki/ApplicationIssues#Application_Working_Directory

Flask session not persisting

Am running with Python 2.7, Apache + mod_wsgi on CentOS 6.3
Things work fine when I am on localhost. However, when I run the code on a vm in Azure, I do not see the session information being persisted across pages.
Basically in my views, I have something like:
#frontend.route('/')
def index():
session['foo'] = 'bar'
print session['foo']
return redirect(url_for("frontend.page2"))
#frontend.route('page2')
def page2():
print session
The print output is:
bar
<SecureCookieSession {}>
My wsgi configuration for apache is:
WSGISocketPrefix /var/run/wsgi
<VirtualHost *:80>
ServerName example.com
ServerAlias example.com
WSGIDaemonProcess myproj threads=5 processes=5
WSGIScriptAlias / /home/mydir/myproj/apache/myproj.wsgi
<Directory /home/mydir/myproj>
WSGIScriptReloading On
WSGIProcessGroup myproj
WSGIApplicationGroup %{GLOBAL}
Order deny,allow
Allow from all
</Directory>
</VirtualHost>
I have the secret_key set:
app.secret_key = os.urandom(24)
I have tried with both setting SERVER_NAME but it doesn't help:
app.config['SERVER_NAME'] = 'example.com'
Any ideas on how I can debug this more?
Thanks!

Don't use app.secret_key = os.urandom(24)!
You're supposed to enter a static value here, not read from os.urandom each time. You've probably misunderstood the example in the docs, it shows you how you can read random data from os.urandom, but it also clearly states:
Just take that thing and copy/paste it into your code and you’re done
If you read it at runtime, then each of your worker processes will have a different secret key! That means if a request is handled by a different worker, the session will break because the cookie is signed with the wrong secret key.

wsgi django not working

im installing django,
the test for wsgi is ok, but when i point my default file to the django test, it doesnt work,
this is the test that works fine:
default: /etc/apache2/sites-available/default
<VirtualHost *:80>
ServerName www.example.com
ServerAlias example.com
ServerAdmin webmaster#example.com
DocumentRoot /var/www
<Directory /var/www/documents>
Order allow,deny
Allow from all
</Directory>
WSGIScriptAlias / /home/ubuntu/djangoProj/micopiloto/application.wsgi
<Directory /home/ubuntu/djangoProj/mysitio/wsgi_handler.py>
Order allow,deny
Allow from all
</Directory>
</VirtualHost>
application.wsgi:: ~/djangoProj/micopiloto
import os
import sys
sys.path.append('/srv/www/cucus/application')
os.environ['PYTHON_EGG_CACHE'] = '/srv/www/cucus/.python-egg'
def application(environ,
start_response):
status = '200 OK'
output = 'Hello World!MK SS9 tkt kkk'
response_headers = [('Content-type', 'text/plain'),
('Content-Length', str(len(output)))]
start_response(status, response_headers)
return [output]
but if I change the default to point to application_sa.wsgi the django test,
it doesnt work :(
application_sa.wsgi
import os, sys
sys.path.append('/home/ubuntu/djangoProj/micopiloto')
os.environ['DJANGO_SETTINGS_MODULE'] = 'micopiloto.settings'
import django.core.handlers.wsgi
application = django.core.handlers.wsgi.WSGIHandler()
I restart the apache server every time i change the wsgi to test,
so what im i missing?
thanks a lot!

Try changing your apache config to the following:
AddHandler wsgi-script .wsgi
WSGISocketPrefix /var/run/wsgi
WSGIRestrictEmbedded On
<VirtualHost *:80>
ServerName www.mydomain.com
WSGIDaemonProcess myprocess processes=2 threads=15
WSGIProcessGroup myprocess
WSGIScriptAlias / /home/ubuntu/djangoProj/micopiloto/application_sa.wsgi
</VirtualHost>
Then add the root folder of your project to sys.path as well in application_sa.wsgi:
sys.path.append('/home/ubuntu/djangoProj')
sys.path.append('/home/ubuntu/djangoProj/micopiloto')