Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 29 Apr 2022 12:15:16 +0000
From:      bugzilla-noreply@freebsd.org
To:        ports-bugs@FreeBSD.org
Subject:   [Bug 263655] www/apache24: Remove automatic mpm fallback configuration
Message-ID:  <bug-263655-7788@https.bugs.freebsd.org/bugzilla/>

next in thread | raw e-mail | index | archive | help
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=3D263655

            Bug ID: 263655
           Summary: www/apache24: Remove automatic mpm fallback
                    configuration
           Product: Ports & Packages
           Version: Latest
          Hardware: Any
                OS: Any
            Status: New
          Severity: Affects Only Me
          Priority: ---
         Component: Individual Port(s)
          Assignee: apache@FreeBSD.org
          Reporter: ports@skyforge.at
          Assignee: apache@FreeBSD.org
             Flags: maintainer-feedback?(apache@FreeBSD.org)

Created attachment 233582
  --> https://bugs.freebsd.org/bugzilla/attachment.cgi?id=3D233582&action=
=3Dedit
www/apache24: Remove automatic mpm fallback configuration

I'd like to propose a patch to remove the automatic fallback generation with
mpm_prefork for www/apache24. Currently the apache24 package greps for a
present MPM configuration in http.conf and creates a fallback configuration=
 in
modules.d, which is included by default. I feel that this is problematic fo=
r a
variety of reasons:


* It doesn't account for modular configurations: If the loading of an MPM
module takes place in an included configuration (as would be the case for m=
ost
modularized configuration schemes, which are increadibly useful and common),
the script generates a fallback configuration that may cause apache to eith=
er
load the wrong mpm module on restart, or cause it to fail starting alltoget=
her.

* It doesn't preserve changes to the file: Even if I keep the
000_mpm_prefork_fallback.conf configuration but uncomment the LoadModule
statement, the configuration is simply overwritten on the next
update/reinstallation of the package.

* From a philosophical perspective, it feels like it's overstepping boundar=
ies:
We nowadays ship with a sample configuration that includes a well-intended
default LoadModule statement for mpm_prefork. Whichever way the user decide=
s to
modify and diverge from that default configuration should be up to them.
Creating non-sample configurations in places that would be sourced by defau=
lt
feels like it's taking things a step too far.

Judging from the commit history, I think there was once a point for this ki=
nd
of behaviour to help with the migration from a static to a dynamic MPM back=
end.
But this was in 2015, so I think it's time to reevaluate this and, perhaps,
move on.

As a minor note: I think the configuration as it stands right now doesn't e=
ven
work properly anymore. At least my apache complains with the following:

Syntax error on line 21 of
/usr/local/etc/apache24/modules.d/000_mpm_prefork_fallback.conf: Cannot load
libexec/apache24/mod_mpm_prefork.so into server: Cannot open
"/usr/local/etc/apache24/libexec/apache24/mod_mpm_prefork.so"

I've attached a patch that (I think?) removes this functionality in its
entirety, as it really shouldn't be necessary anymore. All configurations h=
ad
ample time migrating and new installations come with a sane default anyway.=
 We
could still preserve a warning on install, but I'm really not sure that's
necessary anymore.

In any case it would make my own personal use case a bit nicer to deal with=
, as
I have split the individual modules and their configurations in seperate fi=
les
in modules.d and currently need to pay attention to this everytime I update
apache24, but I'd love to hear feedback and different opinions on this. :)


Cheers,
Sascha

--=20
You are receiving this mail because:
You are the assignee for the bug.=



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-263655-7788>