Fedora 30 with mod_ruid2, aborting processes

I’m seeing issues with httpd + mod_ruid2. To get to the state I’ve done the following on a fresh install of Fedora 30 Server:

# dnf install httpd mod_ruid2
# firewall-cmd --add-service={http,https}
# firewall-cmd --runtime-to-permanent

When curling I get:

# curl -i -v localhost
*   Trying ::1...
* TCP_NODELAY set
* Connected to localhost (::1) port 80 (#0)
> GET / HTTP/1.1
> Host: localhost
> User-Agent: curl/7.64.0
> Accept: */*
> 
* Empty reply from server
* Connection #0 to host localhost left intact
curl: (52) Empty reply from server

I see that children are aborting:

# tail /var/log/httpd/error_log 
[Sun Jul 28 03:03:06.249611 2019] [core:notice] [pid 8702:tid 140545021927744] AH00052: child pid 8808 exit signal Aborted (6)
[Sun Jul 28 03:03:09.253738 2019] [core:notice] [pid 8702:tid 140545021927744] AH00052: child pid 8707 exit signal Aborted (6)
[Sun Jul 28 03:03:09.253869 2019] [core:notice] [pid 8702:tid 140545021927744] AH00052: child pid 8837 exit signal Aborted (6)
[Sun Jul 28 03:03:09.253988 2019] [core:notice] [pid 8702:tid 140545021927744] AH00052: child pid 8838 exit signal Aborted (6)
[Sun Jul 28 03:03:10.256436 2019] [core:notice] [pid 8702:tid 140545021927744] AH00052: child pid 8918 exit signal Aborted (6)
[Sun Jul 28 03:03:11.260291 2019] [core:notice] [pid 8702:tid 140545021927744] AH00052: child pid 8963 exit signal Aborted (6)
[Sun Jul 28 03:03:11.260447 2019] [core:notice] [pid 8702:tid 140545021927744] AH00052: child pid 8964 exit signal Aborted (6)
[Sun Jul 28 03:03:12.261564 2019] [core:notice] [pid 8702:tid 140545021927744] AH00052: child pid 9038 exit signal Aborted (6)
[Sun Jul 28 03:16:47.163868 2019] [core:notice] [pid 8702:tid 140545021927744] AH00052: child pid 9039 exit signal Aborted (6)
[Sun Jul 28 03:16:55.173379 2019] [core:notice] [pid 8702:tid 140545021927744] AH00052: child pid 9040 exit signal Aborted (6)

I’ve looked into this from a few angles, but I’ve not sure how to proceed further in trouble shooting this.

  • Disabled selinux to see if it was a context issue
  • Removed the Fedora mod_ruid2 package and compiled mod_ruid2 from source
  • Switched between prefork, worker, and event. (It should work fine with event, the default)

Maybe someone has encountered this or maybe it’s a bug?
Any help is appreciated, my only ask is that you just don’t suggest php-fpm, there are reasons we aren’t using it.

Hello @igknighted! And welcome to the community :slightly_smiling_face: Please do take a few minutes to go over the introductory posts in #start-here when you have the time. They contain lots of useful information.

Sorry @igknighted, I’m not an expert, and I never used mod_ruid2, but as far as I can see, there are various bug report even on other distros.
In addition on GitHub - mind04/mod-ruid2: mod_ruid2 is a suexec module for apache which takes advantage of POSIX.1e capabilities to increase performance. it seems that the code is a bit unmaintained: latest commit dates to 2013.

After further digging I came across the issues on the repo and there’s some bug reports on debian for it as well. On one of the debian bug reports someone mentioned that issues were happening around setgroups().

I’m starting to suspect that something changed in kernel space that broke this. Which is way outside of my wheelhouse. So I guess I’ll just have to start going through the ruid2 code and learning it, to debug it.