Gateway Timeout caused by "Failure to configure the daemon process"

[Russell Parker]

Hi all,

I've been banging my head against the wall trying to fix this. This issue seems similar but I haven't been able to wrangle a solution out of it. 

• mod_wsgi 4.3
• Apache 2.4
• Debian Jessie
  
• Python 2.7.9

I'm trying to setup a Docker image running a Flask app and I've gotten to a point where I can run the Flask local development server and retrieve (pretty simple, templated html) pages. When I try to run it on Apache though, there is a gateway timeout and I find the following error in my VirutalHost log file:

 [wsgi:error] ... Timeout when reading response headers from daemon process 'molmod': /var/www/app/molmod.wsgi

In the genreal Apache log file there is the following error over and over:

[wsgi:alert] ...  mod_wsgi (pid=<x>): Failure to configure the daemon process correctly and process left in unspecified state. Restarting daemon process after delay.

It seems like the daemon gets stuck restarting over and over instead of serving the Flask site. 

Here is my config file:

ErrorLog logs/error_log

WSGISocketPrefix /var/run
# turn off initialization of Python interpreter

WSGIRestrictEmbedded On

<VirtualHost *:80>

    Alias /static/ /var/www/app/static/

    <Directory /var/www/app/static/>

        Require all granted

    </Directory>

    WSGIDaemonProcess molmod user=flask_user group=flask_group threads=5

    WSGIScriptAlias / /var/www/app/molmod.wsgi

    <Directory /var/www/app>

        WSGIProcessGroup molmod

        WSGIApplicationGroup %{GLOBAL}

        Require all granted

    </Directory>

    LogLevel info

    ErrorLog logs/flask_log

</VirtualHost>

If it helps at all, here is the Dockerfile I am using as well:

FROM httpd:2.4

RUN apt-get -y update && \

    apt-get install -y libapache2-mod-wsgi

RUN useradd flask_user && groupadd flask_group

COPY ./molmod /var/www/app

RUN apt-get install -y python-pip && \

    pip install -r /var/www/app/requirements.txt

COPY ./httpd.conf /tmp/httpd.conf

RUN echo "LoadModule wsgi_module /usr/lib/apache2/modules/mod_wsgi.so" >> \

    "$HTTPD_PREFIX"/conf/httpd.conf

RUN cat /tmp/httpd.conf >> "$HTTPD_PREFIX"/conf/httpd.conf && \

    rm /tmp/httpd.conf

EXPOSE 80

CMD ["httpd-foreground"]

Thanks in advance!

[Graham Dumpleton]

On 30 Jun 2016, at 7:54 PM, Russell Parker <russell...@gmail.com> wrote:

Hi all,

I've been banging my head against the wall trying to fix this. This issue seems similar but I haven't been able to wrangle a solution out of it. 

• mod_wsgi 4.3
• Apache 2.4
• Debian Jessie
  
• Python 2.7.9

I'm trying to setup a Docker image running a Flask app and I've gotten to a point where I can run the Flask local development server and retrieve (pretty simple, templated html) pages. When I try to run it on Apache though, there is a gateway timeout and I find the following error in my VirutalHost log file:

 [wsgi:error] ... Timeout when reading response headers from daemon process 'molmod': /var/www/app/molmod.wsgi

In the genreal Apache log file there is the following error over and over:

[wsgi:alert] ...  mod_wsgi (pid=<x>): Failure to configure the daemon process correctly and process left in unspecified state. Restarting daemon process after delay.

This would usually indicate you are trying to set the user/group of the mod_wsgi daemon process with a user or group that doesn’t exist inside of the Docker image.

I seriously would advise against you trying to construct your own Docker image. Creating Docker images properly that follow best practices, especially if trying to run Apache is non trivial.

For now I would suggest you use my base image:

* https://github.com/GrahamDumpleton/mod_wsgi-docker

I have an even better base image than this one which can do a lot more and security wise is much better, but not yet ready to let people use the newer one. So try the grahamdumpleton/mod-wsgi-docker ones first.

If you think that creating Docker images is trivial, you might want to read some of my blog posts about some of the issues you need to deal with.

* http://blog.dscpl.com.au/2016/01/roundup-of-docker-issues-when-hosting.html

The image above I reference doesn’t adhere to all the things one should do, but the newer one will and more.

Graham

--
You received this message because you are subscribed to the Google Groups "modwsgi" group.
To unsubscribe from this group and stop receiving emails from it, send an email to modwsgi+u...@googlegroups.com.
To post to this group, send email to mod...@googlegroups.com.
Visit this group at https://groups.google.com/group/modwsgi.
For more options, visit https://groups.google.com/d/optout.

[Russell Parker]

Regarding setting the image to a user or group that doesn't exist I expected this Dockerfile to set things up enough

RUN useradd flask_user && groupadd flask_group

since the same user name and group name are referenced in the httpd.conf: 

WSGIDaemonProcess molmod user=flask_user group=flask_group threads=5

A similar setup worked earlier when doing a test on a CentOS box.


Regardless, I will try using your Docker image instead and look forward to the new more secure version.

Thanks for taking the time to help!