503 Service Unavilable Error - Daemon mode

[Guddu]

After configuring daemon mode, I started getting a 503 Server Unavailable error.

There are lot of posts on stackoverflow and a dedicate section in the troubleshooting page at https://modwsgi.readthedocs.io/en/develop/user-guides/frequently-asked-questions.html and https://modwsgi.readthedocs.io/en/develop/user-guides/configuration-issues.html but none of the suggestions seem to be helping me at the moment. Appreciate your guidance.

The error logs have the following

[Wed Mar 18 09:30:17.649428 2020] [wsgi:error] [pid 14044] (13)Permission denied: [client 10.52.7.18:2505] mod_wsgi (pid=14044): Unable to connect to WSGI daemon process 'app_qa' on '/run/httpd/wsgi/.14039.0.1.sock' as user with uid=1001.

The directory in question has read and write permission to app_user

The httpd processes start fine. Below is the ps output

root     14197     1  0 09:30 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND
app_user  14202 14197  0 09:30 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND
app_user  14203 14197  0 09:30 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND
app_user  14204 14197  0 09:30 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND
app_user  14205 14197  0 09:30 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND
app_user  14206 14197  0 09:30 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND

 

The socket flies get written

 

[root@server wsgi]# ls -ltra
total 0
srwx------ 1 app_user root      0 Mar 18 09:30 .14197.0.1.sock
srwx------ 1 app_user root      0 Mar 18 09:30 .14197.0.2.sock

My apache build info is as follows

[app_user@server app_qa]$ httpd -V
AH00558: httpd: Could not reliably determine the server's fully qualified domain name, using 10.21.2.136. Set the 'ServerName' directive globally to suppress this message
Server version: Apache/2.4.6 (Red Hat Enterprise Linux)
Server built:   Jun  9 2019 13:01:04
Server's Module Magic Number: 20120211:24
Server loaded:  APR 1.4.8, APR-UTIL 1.5.2
Compiled using: APR 1.4.8, APR-UTIL 1.5.2
Architecture:   64-bit
Server MPM:     prefork
  threaded:     no
    forked:     yes (variable process count)
Server compiled with....
 -D APR_HAS_SENDFILE
 -D APR_HAS_MMAP
 -D APR_HAVE_IPV6 (IPv4-mapped addresses enabled)
 -D APR_USE_SYSVSEM_SERIALIZE
 -D APR_USE_PTHREAD_SERIALIZE
 -D SINGLE_LISTEN_UNSERIALIZED_ACCEPT
 -D APR_HAS_OTHER_CHILD
 -D AP_HAVE_RELIABLE_PIPED_LOGS
 -D DYNAMIC_MODULE_LIMIT=256
 -D HTTPD_ROOT="/etc/httpd"
 -D SUEXEC_BIN="/usr/sbin/suexec"
 -D DEFAULT_PIDLOG="/run/httpd/httpd.pid"
 -D DEFAULT_SCOREBOARD="logs/apache_runtime_status"
 -D DEFAULT_ERRORLOG="logs/error_log"
 -D AP_TYPES_CONFIG_FILE="conf/mime.types"
 -D SERVER_CONFIG_FILE="conf/httpd.conf"

 [app_user@server app_qa]$ httpd -l
Compiled in modules:
  core.c
  mod_so.c
  http_core.c

My httpd conf is as follows
 

WSGISocketPrefix /run/httpd/wsgi/
<VirtualHost *:8999>
WSGIDaemonProcess app_qa processes=2 display-name=%{GROUP} python-path=/opt/app/app_qa  socket-user=app_user
WSGIProcessGroup app_qa
WSGIScriptAlias /app_qa /opt/app/app_qa/app/wsgi.py  process-group=app_qa
Alias /static_app_qa /opt/app/app_qa/staticfiles

<Directory /opt/app/app_qa/app/>
        <Files wsgi.py>
                Require all granted
        </Files>
</Directory>

<Directory /opt/app/app_qa/staticfiles/>
        Require all granted
</Directory>
</VirtualHost>

Appreciate your help on this matter.

[Graham Dumpleton]

Why are you setting 'socket-user'?

That is only required when using Apache PrivilegesMode as SECURE, or using certain Apache MPMs.

What happens when you don't use that?

I want to see the original error messages and permissions on both the socket files, and the directory the sockets are in, before changing any configuration from what one would normally use.

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 view this discussion on the web visit https://groups.google.com/d/msgid/modwsgi/82ab46fe-adc1-42fd-90ad-abaf05f58613%40googlegroups.com.

[Guddu]

Dear Graham, Thanks for your response.

I removed socket-user=app_user and now I see a different behaviour. One of the app work while the other does not. Both are exactly identical one is meant for QA and other for PROD and the QA URL works while for the PROD URL it shows an error

Not Found The requested URL /app_prod was not found on this server.

The   QA    URL is http://<ip>:8999/app_qa

The PROD URL is http://<ip>:8999/app_prod

The initial reason why I had to move to daemon mode is to be able to set a python-path for both of them separately as WSGIPythonPath cannot be used inside <VirtualHost>

This is my httpd.conf setting.

WSGISocketPrefix /run/httpd/wsgi/
<VirtualHost *:8999>

WSGIDaemonProcess app_qa processes=1 display-name=%{GROUP} python-path=/opt/app/app_qa/app

WSGIProcessGroup app_qa
WSGIScriptAlias /app_qa /opt/app/app_qa/app/wsgi.py  process-group=app_qa
Alias /static_app_qa /opt/app/app_qa/staticfiles

<Directory /opt/app/app_qa/app/>
        <Files wsgi.py>
                Require all granted
        </Files>
</Directory>

<Directory /opt/app/app_qa/staticfiles/>
        Require all granted
</Directory>
</VirtualHost>

<VirtualHost *:8999>
WSGIDaemonProcess app_prod processes=1 display-name=%{GROUP} python-path=/opt/app/app_prod/app
WSGIProcessGroup app_prod
WSGIScriptAlias /app_prod /opt/app/app_prod/app/wsgi.py  process-group=app_prod
Alias /static_app_prod /opt/app/app_prod/staticfiles

<Directory /opt/app/app_prod/app/>

        <Files wsgi.py>
                Require all granted
        </Files>
</Directory>

<Directory /opt/app/app_prod/staticfiles/>

        Require all granted
</Directory>
</VirtualHost>

The netstat output is as follows indicating that the socket files were created in the desired folder. The permission on files and directories also follows

[root@server app]# netstat -an | grep http
unix  2      [ ACC ]     STREAM     LISTENING     1107556  /run/httpd/wsgi/.13750.0.1.sock
unix  2      [ ACC ]     STREAM     LISTENING     1107558  /run/httpd/wsgi/.13750.0.2.sock

[root@server app]# ls -ltr /run/httpd/wsgi/.13750.0.1.sock /run/httpd/wsgi/.13750.0.2.sock
srwx------ 1 app_user root 0 Mar 18 19:29 /run/httpd/wsgi/.13750.0.2.sock
srwx------ 1 app_user root 0 Mar 18 19:29 /run/httpd/wsgi/.13750.0.1.sock

[root@server app]# ls -ltr /run/httpd/
total 8
drwx------ 2 apache  apache  40 Mar 10 13:50 htcacheclean
-rw-r--r-- 1 root    root     8 Mar 18 19:29 authdigest_shm.13750
-rw-r--r-- 1 root    root     6 Mar 18 19:29 httpd.pid
drwxr-xr-x 2 app_user app_user 80 Mar 18 19:29 wsgi

[root@server app]# ls -ltr /run | tail -2
-rw-rw-r--  1 root     utmp     3072 Mar 18 19:11 utmp
drwx--x---  4 root     apache    120 Mar 18 19:29 httpd

Also the ps -ef | grep httpd output is

root     13750     1  0 19:29 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND
app_user  13753 13750  0 19:29 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND
app_user  13754 13750  0 19:29 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND
app_user  13755 13750  0 19:29 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND
app_user  13756 13750  0 19:29 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND
app_user  13757 13750  0 19:29 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND

I am a bit confused as everything seems alright.

Appreciate your help

On Wednesday, 18 March 2020 18:57:26 UTC-3, Graham Dumpleton wrote:

Why are you setting 'socket-user'?

That is only required when using Apache PrivilegesMode as SECURE, or using certain Apache MPMs.

What happens when you don't use that?

I want to see the original error messages and permissions on both the socket files, and the directory the sockets are in, before changing any configuration from what one would normally use.

Graham

To unsubscribe from this group and stop receiving emails from it, send an email to mod...@googlegroups.com.

[Graham Dumpleton]

You can't have two VirtualHost definitions for the same port number unless they have ServerName set for each to a different hostname. Even then, you can't access them by IP and have them distinguished, you would need to use the respective hostnames.

So if you are going to use an IP address only, have one VirtualHost only and use inside of that for mod_wsgi:

WSGIDaemonProcess app_qa processes=1 display-name=%{GROUP} python-path=/opt/app/app_qa/app

WSGIScriptAlias /app_qa /opt/app/app_qa/app/wsgi.py  process-group=app_qa application-group=%{GLOBAL}

WSGIDaemonProcess app_prod processes=1 display-name=%{GROUP} python-path=/opt/app/app_prod/app

WSGIScriptAlias /app_prod /opt/app/app_prod/app/wsgi.py  process-group=app_prod application-group=%{GLOBAL}

Add your other stuff for Alias etc as well for each.

To unsubscribe from this group and stop receiving emails from it, send an email to modwsgi+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/modwsgi/21647165-18ba-4f06-92d0-b224f06e8311%40googlegroups.com.

[Guddu]

Dear Graham, so i changed it to the following bringing them under one single VirtualHost and added application-group=%{GLOBAL} to WSGIScriptAlias but the same situation persists. I can access only the first app not the second one. Here is the .conf setting.

<VirtualHost *:8999>

WSGIDaemonProcess app_qa processes=1 display-name=%{GROUP} python-path=/opt/app/app_qa/app
WSGIProcessGroup app_qa

WSGIScriptAlias /app_qa /opt/app/app_qa/app/wsgi.py  process-group=app_qa application-group=%{GLOBAL}
Alias /static_app_qa /opt/app/app_qa/staticfiles

<Directory /opt/app/app_qa/app/>
        <Files wsgi.py>
                Require all granted
        </Files>
</Directory>


<Directory /opt/app/app_qa/staticfiles/>
        Require all granted
</Directory>

WSGIDaemonProcess app_prod processes=1 display-name=%{GROUP} python-path=/opt/app/app_prod/app
WSGIProcessGroup app_prod

WSGIScriptAlias /app_prod /opt/app/app_prod/app/wsgi.py  process-group=app_prod application-group=%{GLOBAL}
Alias /static_app_prod /opt/app/app_prod/staticfiles

<Directory /opt/app/app_prod/app/>
        <Files wsgi.py>
                Require all granted
        </Files>
</Directory>


<Directory /opt/app/app_prod/staticfiles/>
        Require all granted
</Directory>
</VirtualHost>

Kindly guide.

On Wednesday, 18 March 2020 21:46:35 UTC-3, Graham Dumpleton wrote:

You can't have two VirtualHost definitions for the same port number unless they have ServerName set for each to a different hostname. Even then, you can't access them by IP and have them distinguished, you would need to use the respective hostnames.

So if you are going to use an IP address only, have one VirtualHost only and use inside of that for mod_wsgi:

WSGIDaemonProcess app_qa processes=1 display-name=%{GROUP} python-path=/opt/app/app_qa/app

WSGIScriptAlias /app_qa /opt/app/app_qa/app/wsgi.py  process-group=app_qa application-group=%{GLOBAL}

WSGIDaemonProcess app_prod processes=1 display-name=%{GROUP} python-path=/opt/app/app_prod/app
WSGIScriptAlias /app_prod /opt/app/app_prod/app/wsgi.py  process-group=app_prod application-group=%{GLOBAL}

Add your other stuff for Alias etc as well for each.

To view this discussion on the web visit https://groups.google.com/d/msgid/modwsgi/21647165-18ba-4f06-92d0-b224f06e8311%40googlegroups.com.

[Graham Dumpleton]

That config looks fine. Are you absolutely sure Apache was restarted correctly? Check the Apache error logs to make sure there wasn't a config error that prevent a restart to use the new config.

To unsubscribe from this group and stop receiving emails from it, send an email to modwsgi+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/modwsgi/d79bd082-9468-4a1c-bf86-8e377718ef48%40googlegroups.com.

[Guddu]

Dear Graham. Yes. The restart went well. Here is the log

[root@server app]# ps -ef| grep http
root     15399     1  0 19:55 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND
app_user  15402 15399  0 19:55 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND
app_user  15403 15399  0 19:55 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND
app_user  15404 15399  0 19:55 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND
app_user  15405 15399  0 19:55 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND
app_user  15406 15399  0 19:55 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND
[root@server app]# date
Wed Mar 18 20:08:06 -05 2020
[root@server app]# service httpd restart
Redirecting to /bin/systemctl restart httpd.service
[root@server app]# ps -ef| grep http
root     16129     1  3 20:08 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND
app_user  16132 16129  1 20:08 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND
app_user  16133 16129  1 20:08 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND
app_user  16134 16129  1 20:08 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND
app_user  16135 16129  1 20:08 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND
app_user  16136 16129  1 20:08 ?        00:00:00 /usr/sbin/httpd -DFOREGROUND

On Wednesday, 18 March 2020 22:00:30 UTC-3, Graham Dumpleton wrote:

That config looks fine. Are you absolutely sure Apache was restarted correctly? Check the Apache error logs to make sure there wasn't a config error that prevent a restart to use the new config.

To view this discussion on the web visit https://groups.google.com/d/msgid/modwsgi/d79bd082-9468-4a1c-bf86-8e377718ef48%40googlegroups.com.

[Graham Dumpleton]

Add a syntax error to the file. Using "XXX" on a line by itself is enough and restart Apache again. See if it fails to restart because of the error. This will check whether the file is being read. If this is in a file in a sites-available directory, ensure that the site has been enabled and the sites-enabled directory contains a symlink to the file in the sites-available directory and is not a separate file copy. Also ensure you don't have a VirtualHost for the same port elsewhere in any config file, and that you don't have a VirtualHost which uses "*" for the port. Also try commenting out the config for the QA site and see if anything changes.

So can only suggest you double check everything.

To unsubscribe from this group and stop receiving emails from it, send an email to modwsgi+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/modwsgi/5cb92649-cb87-4d82-a74f-5aebeac2f90c%40googlegroups.com.

[Guddu]

Dear Graham, adding an XXX and restarting failed as expected.

There is no sites-available directory and no other VirtualHosts for the same port or for *.

I removed the QA Site config leaving only the PROD one and it gives the same error which is weird. I have checked all paths and they are fine. Kindly guide. This is the current minimal configuration that I have. It has to be something sillly now which I can't seem to figure out.

WSGISocketPrefix /run/httpd/wsgi/

<VirtualHost *:8999>

WSGIDaemonProcess app_prod processes=1 display-name=%{GROUP} python-path=/opt/app/app_prod/app
WSGIProcessGroup app_prod
WSGIScriptAlias /app_prod /opt/app/app_prod/app/wsgi.py  process-group=app_prod application-group=%{GLOBAL}
Alias /static_app_prod /opt/app/app_prod/staticfiles

<Directory /opt/app/app_prod/app/>
        <Files wsgi.py>
                Require all granted
        </Files>
</Directory>

<Directory /opt/app/app_prod/staticfiles/>
        Require all granted
</Directory>
</VirtualHost>

On Wednesday, 18 March 2020 22:20:07 UTC-3, Graham Dumpleton wrote:

Add a syntax error to the file. Using "XXX" on a line by itself is enough and restart Apache again. See if it fails to restart because of the error. This will check whether the file is being read. If this is in a file in a sites-available directory, ensure that the site has been enabled and the sites-enabled directory contains a symlink to the file in the sites-available directory and is not a separate file copy. Also ensure you don't have a VirtualHost for the same port elsewhere in any config file, and that you don't have a VirtualHost which uses "*" for the port. Also try commenting out the config for the QA site and see if anything changes.

So can only suggest you double check everything.

To unsubscribe from this group and stop receiving emails from it, send an email to mod...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/modwsgi/5cb92649-cb87-4d82-a74f-5aebeac2f90c%40googlegroups.com.

[Graham Dumpleton]

Did the QA URL still work after you removed it?

You sure you aren't using a standalone WSGI server run from command line on port 8999 as well, preventing Apache from using the port?

Does the file '/opt/app/app_prod/app/wsgi.py' exist?

To unsubscribe from this group and stop receiving emails from it, send an email to modwsgi+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/modwsgi/61105e58-3f62-409d-a224-afa986c769c5%40googlegroups.com.

[Guddu]

Dear Graham, the wsgi.py file exists.

[root@server app]# ls -ltr /opt/app/app_prod/app/wsgi.py
-rw-r--r-- 1 app_user app_user 411 Mar 18 06:53 /opt/app/app_prod/app/wsgi.py

There is no other entity using the port 8999

[root@server app]# netstat -an | grep 8999
tcp6       0      0 :::8999                 :::*                    LISTEN
[root@server app]# service httpd stop
Redirecting to /bin/systemctl stop httpd.service
[root@server app]# netstat -an | grep 8999
[root@server app]#

If i put in the QA wsgi app it works. Below is the config.

WSGISocketPrefix /run/httpd/wsgi/
<VirtualHost *:8999>
WSGIDaemonProcess app_qa processes=1 display-name=%{GROUP} python-path=/opt/app/app_qa/app

#WSGIProcessGroup app_qa

WSGIScriptAlias /app_qa /opt/app/app_qa/app/wsgi.py  process-group=app_qa application-group=%{GLOBAL}
Alias /static_app_qa /opt/app/app_qa/staticfiles

<Directory /opt/app/app_qa/app/>
        <Files wsgi.py>
                Require all granted
        </Files>
</Directory>

<Directory /opt/app/app_qa/staticfiles/>
        Require all granted
<

/Directory>

</VirtualHost>

If i put in the PROD wsgi app it does not works Below is the config.

WSGISocketPrefix /run/httpd/wsgi/

<VirtualHost *:8999>
WSGIDaemonProcess app_prod processes=1 display-name=%{GROUP} python-path=/opt/app/app_prod/app
WSGIProcessGroup app_prod
WSGIScriptAlias /app_prod /opt/app/app_prod/app/wsgi.py  process-group=app_prod application-group=%{GLOBAL}
Alias /static_app_prod /opt/app/app_prod/staticfiles

<Directory /opt/app/app_prod/app/>
        <Files wsgi.py>
                Require all granted
        </Files>
</Directory>

<Directory /opt/app/app_prod/staticfiles/>
        Require all granted
</Directory>
</VirtualHost>

On Wednesday, 18 March 2020 22:51:56 UTC-3, Graham Dumpleton wrote:

Did the QA URL still work after you removed it?

You sure you aren't using a standalone WSGI server run from command line on port 8999 as well, preventing Apache from using the port?

Does the file '/opt/app/app_prod/app/wsgi.py' exist?

To unsubscribe from this group and stop receiving emails from it, send an email to mod...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/modwsgi/61105e58-3f62-409d-a224-afa986c769c5%40googlegroups.com.

[Graham Dumpleton]

But when you took the QA app out, the QA app definitely didn't work if you tried accessing it? Just trying to make sure that the QA config isn't being read from somewhere else.

Also make sure you have no strange unicode characters in the URL path of WSGIScriptAlias. Type it in again from scratch.

To unsubscribe from this group and stop receiving emails from it, send an email to modwsgi+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/modwsgi/707a3fe4-b289-4f8d-9166-d57c5687ceb9%40googlegroups.com.

[Guddu]

Dear Graham, after taking QA out and accessing it it did not work.

So I started the whole server just to see if that helps but strangely I am back to the original error.

Here is the message in the apache logs upon accessing both the apps in turn one after the other

[Wed Mar 18 21:17:00.043635 2020] [wsgi:error] [pid 19948] (13)Permission denied: [client 10.52.7.170:1251] mod_wsgi (pid=19948): Unable to connect to WSGI daemon process 'app_qa' on '/run/httpd/wsgi/.19945.0.1.sock' as user with uid=1001.[Wed Mar 18 21:18:21.667963 2020] [wsgi:error] [pid 19949] (13)Permission denied: [client 10.52.7.170:1254] mod_wsgi (pid=19949): Unable to connect to WSGI daemon process 'app_prod' on '/run/httpd/wsgi/.19945.0.2.sock' as user with uid=1001.

That uid of 1001 corresponds to the same user with which i am runing the apache and specified in User and Groups directive.

The permissions on those socket files are as follows

[root@server app]# ls -ltra /run/httpd/wsgi/
total 0
srwx------ 1 app_user root      0 Mar 18 21:14 .19945.0.1.sock
srwx------ 1 app_user root      0 Mar 18 21:14 .19945.0.2.sock

On Wednesday, 18 March 2020 23:03:49 UTC-3, Graham Dumpleton wrote:

But when you took the QA app out, the QA app definitely didn't work if you tried accessing it? Just trying to make sure that the QA config isn't being read from somewhere else.

Also make sure you have no strange unicode characters in the URL path of WSGIScriptAlias. Type it in again from scratch.

To unsubscribe from this group and stop receiving emails from it, send an email to mod...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/modwsgi/707a3fe4-b289-4f8d-9166-d57c5687ceb9%40googlegroups.com.

[Graham Dumpleton]

Get rid the trailing slash on:

WSGISocketPrefix /run/httpd/wsgi/

it should be:

WSGISocketPrefix /run/httpd/wsgi/

Remove the directory:

/run/httpd/wsgi

    

What are the permissions on the directory '/run/httpd'. Ie.,

ls -lasd /run/httpd

There is no reason to change User/Group directives for Apache as a whole.

Graham

To unsubscribe from this group and stop receiving emails from it, send an email to modwsgi+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/modwsgi/be886f37-3e60-4b85-8f0a-971c6bcd8e4c%40googlegroups.com.

[Graham Dumpleton]

On 19 Mar 2020, at 1:34 pm, Graham Dumpleton <graham.d...@gmail.com> wrote:

Get rid the trailing slash on:

WSGISocketPrefix /run/httpd/wsgi/

it should be:

WSGISocketPrefix /run/httpd/wsgi/

Whoops.

WSGISocketPrefix /run/httpd/wsgi

[Guddu]

Dear Graham, Same error after removing the trailing slash. The socket files get created anyways in /run/httpd/wsgi.21949.0.1.sock

I believe we are just one step away from solving this but I am a bit lost.

On Wednesday, 18 March 2020 23:35:22 UTC-3, Graham Dumpleton wrote:

On 19 Mar 2020, at 1:34 pm, Graham Dumpleton <graham....@gmail.com> wrote:

Get rid the trailing slash on:

WSGISocketPrefix /run/httpd/wsgi/

it should be:

WSGISocketPrefix /run/httpd/wsgi/

Whoops.

WSGISocketPrefix /run/httpd/wsgi

Remove the directory:

/run/httpd/wsgi

    

What are the permissions on the directory '/run/httpd'. Ie.,

ls -lasd /run/httpd

There is no reason to change User/Group directives for Apache as a whole.

Graham

To unsubscribe from this group and stop receiving emails from it, send an email to mod...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/modwsgi/be886f37-3e60-4b85-8f0a-971c6bcd8e4c%40googlegroups.com.

[Graham Dumpleton]

I have asked multiple times now what are the permissions on the socket directory, which should now be '/run/httpd'. Ie.,

ls -lasd /run/httpd

If the directory is not readable by others, it will not work.

Try with:

   WSGISocketPrefix /tmp

temporarily if have to:

https://modwsgi.readthedocs.io/en/develop/user-guides/configuration-issues.html

To unsubscribe from this group and stop receiving emails from it, send an email to modwsgi+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/modwsgi/2dc3154b-d634-464d-aa3a-39c33b394057%40googlegroups.com.

[Guddu]

Dear Graham, Taking it to /tmp solved it.

So I looked and set the /run/httpd directory permission to the following changing it from root owner to app_user owner and that also solves the issue.

Thanks a lot for your help on this and your time.I can't express how much this has helped me today.

[root@server run]# ls -lasd /run/httpd
0 drwxr-xr-x 2 app_user app_user 120 Mar 18 22:00 /run/httpd

[root@server run]# ls -lasd /run/httpd/*
4 -rw-r--r-- 1 root    root 8 Mar 18 22:00 /run/httpd/authdigest_shm.22532
4 -rw-r--r-- 1 root    root 6 Mar 18 22:00 /run/httpd/httpd.pid
0 srwx------ 1 app_user root 0 Mar 18 22:00 /run/httpd/wsgi.22532.0.1.sock
0 srwx------ 1 app_user root 0 Mar 18 22:00 /run/httpd/wsgi.22532.0.2.sock

On Wednesday, 18 March 2020 23:55:09 UTC-3, Graham Dumpleton wrote:

I have asked multiple times now what are the permissions on the socket directory, which should now be '/run/httpd'. Ie.,

ls -lasd /run/httpd

If the directory is not readable by others, it will not work.

Try with:

   WSGISocketPrefix /tmp

temporarily if have to:

https://modwsgi.readthedocs.io/en/develop/user-guides/configuration-issues.html

To unsubscribe from this group and stop receiving emails from it, send an email to mod...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/modwsgi/2dc3154b-d634-464d-aa3a-39c33b394057%40googlegroups.com.

[Graham Dumpleton]

At some point in time you changed owner/group of the directory '/run/httpd', and possibly permissions as well. That is something that shouldn't have been changed. You should work out what original ownership and permissions are and restore it.

So since /tmp works, follow the guidelines in:

https://modwsgi.readthedocs.io/en/develop/user-guides/configuration-issues.html#location-of-unix-sockets

about using:

WSGISocketPrefix /var/run/wsgi

To unsubscribe from this group and stop receiving emails from it, send an email to modwsgi+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/modwsgi/352a4909-7703-4b6e-aa4d-a5f0ef807fe0%40googlegroups.com.