You must have an old installation. At some point gpg-agent.service was enabled globally by default, but it's not the case anymore since https://svnweb.mageia.org/packages?view=revision&revision=1654247.

You can use the following to reset to the current default. It removes "/etc/systemd/user/default.target.wants/gpg-agent.service" symlink.

# systemctl --no-reload preset --global gpg-agent

CC: (none) => jani.valimaa

Since https://svnweb.mageia.org/packages?view=revision&revision=1654247 gpg-agent is socket activated, and enabling it is up to user.

$ systemctl enable --user gpg-agent.socket

Globally gpg-agent can be enabled for all users with

# systemctl enable --global gpg-agent.socket

sure. This is a server - it gets updates. I'm not sure, if it was created with mandriva or it was renewed when mageia was found.

But the change was only 3 years ago - most of my systems are way older.
The rpm should have removed the global systemd file.

Thanks again Jani for your expert explanations.

@Marc: wha

CC: (none) => lewyssmith

@Marc: what package are we talking about? These look the most likely:
 lib64gpg-error0
 lib64gpgme11
 lib64gpgmepp6
 lib64qgpgme15
There is no gpg-agent.

@Jani: "The rpm should have removed the global systemd file"
Is this possible & reasonable?

It is gnupg2:/usr/bin/gpg-agent

The installation of gnupg2 in mga7/8 ? registered the service as global. So each user session started the gpg-agent.
As it was decided to add the registration in user via sockets, the old file should have been removed. I guess other people which only do updates will have the service beeing started every time, and maybe having some wired trouble because of it. (e.g. most backup programs require gpg)

(In reply to Lewis Smith from comment #5)
> @Jani: "The rpm should have removed the global systemd file"
> Is this possible & reasonable?

It's possible, for example, with %triggerpostun scriptlet.
https://svnweb.mageia.org/packages?view=revision&revision=1700387

It was removed in next revision. Unfortunately I can't remember the reason for it.
https://svnweb.mageia.org/packages?view=revision&revision=1700388

The origin:
Revision 1654247 Mon Dec 7 16:54:56 2020 joequant
"move activation to user to avoid needing systemd"
which looks to be what Marc is asking for.

The first change noted above is commented:
Revision 1700387 Sun Mar 7 11:39:04 2021
"reset global gpg-agent user service setting to system preset (disable) after change in r1654247"

and the second above:
Revision 1700388 Sun Mar 7 11:42:22 2021
"revert r1700387 to not force changes in users' systems (SILENT)"
that is, the previous change was immediately reverted.

So what do we ask the packagers to do? I prefer to forward this with a +ve recommendation.

I have no idea what is 'a +ve recommendation'. Anyhow I guess we want to force the change/removal anyway as otherwise there can be problems.

Comment just says "revert r1700387 to not force changes in users' systems"
which I usally agree, but it was set by the package; so resetting a dumb decission should be ok. And it makes trouble.

So, if you're ok with it; I will just (re-)add the patch and build the package for mga9

I don't think the famous nobody, gnupg2's maintainer, will object.

Suggested advisory:
========================
Updated gnupg2 package to remove gpg-agent startup on each user session (even for ssh-root sessions).
This behaviour was introduced by an older version of gnupg2 and is now corrected. The old behaviour can cause global services to be restarted (e.g. databases, webserver) by systemd, when gnupg2 crashes.


Updated packages in core/updates_testing:
========================
gnupg2-debugsource-2.3.8-1.1.mga9
gnupg2-2.3.8-1.1.mga9
gnupg2-debuginfo-2.3.8-1.1.mga9

SRPM:
gnupg2-2.3.8-1.1.mga9.src.rpm

Assignee: bugsquad => qa-bugs

I had to fix %%triggerpostun scriptlet version to trigger also with latest pkgs in mga9 and cauldron.

Updated RPMS/SRPMS to test in mga9 core/updates_testing:
gnupg2-2.3.8-1.2.mga10

Isn't it better to just trigger it everytime? I was not familiar with that syntax on triggerpostun

RH mageia 9 x86_64

LC_ALL=C urpmi --auto --auto-update 
medium "QA Testing (32-bit)" is up-to-date
medium "QA Testing (64-bit)" is up-to-date
medium "Core Release (distrib1)" is up-to-date
medium "Core Updates (distrib3)" is up-to-date
medium "Nonfree Release (distrib11)" is up-to-date
medium "Nonfree Updates (distrib13)" is up-to-date
medium "Tainted Release (distrib21)" is up-to-date
medium "Tainted Updates (distrib23)" is up-to-date
medium "Core 32bit Release (distrib31)" is up-to-date
medium "Core 32bit Updates (distrib32)" is up-to-date
medium "Nonfree 32bit Release (distrib36)" is up-to-date
medium "Tainted 32bit Release (distrib41)" is up-to-date
medium "Tainted 32bit Updates (distrib42)" is up-to-date

installing gnupg2-2.3.8-1.2.mga9.x86_64.rpm from //home/katnatek/qa-testing/x86_64
Preparing...                     ##################################################################################################
      1/1: gnupg2                ##################################################################################################
      1/1: removing gnupg2-2.3.8-1.mga9.x86_64
                                 ##################################################################################################

gpg --list-keys
gpg --list-secret-keys

Shows me the same information produced before the update (information about the key used to sign packages in blogdrake's repository)
sign a few packages without issue
Encrypt text file with gpg --symmetric 
Decrypt the file with gpg --decrypt file.gpg > newfile
diff -s original newfile says the files are identical

OK for me for functionalities not related to the bug and tested in previous rounds
If Marc or Jani can confirm that fix the issue, I'll put the OK

To properly test the upgrade symlink /etc/systemd/user/default.target.wants/gpg-agent.service -> /usr/lib/systemd/user/gpg-agent.service must exist.

Steps:
# mkdir -p /etc/systemd/user/default.target.wants/
# ln -sf /usr/lib/systemd/user/gpg-agent.service /etc/systemd/user/default.target.wants/gpg-agent.service
# urpmi gnupg2
installing gnupg2-2.3.8-1.2.mga9.x86_64.rpm from /var/cache/urpmi/rpms         
Preparing...                     #############################################
      1/1: gnupg2                #############################################
      1/1: removing gnupg2-2.3.8-1.mga9.x86_64
                                 #############################################
Removed "/etc/systemd/user/default.target.wants/gpg-agent.service".

(In reply to Jani Välimaa from comment #16)
> To properly test the upgrade symlink
> /etc/systemd/user/default.target.wants/gpg-agent.service ->
> /usr/lib/systemd/user/gpg-agent.service must exist.
> 
> Steps:
> # mkdir -p /etc/systemd/user/default.target.wants/
> # ln -sf /usr/lib/systemd/user/gpg-agent.service
> /etc/systemd/user/default.target.wants/gpg-agent.service
> # urpmi gnupg2
> installing gnupg2-2.3.8-1.2.mga9.x86_64.rpm from /var/cache/urpmi/rpms      
> 
> Preparing...                    
> #############################################
>       1/1: gnupg2               
> #############################################
>       1/1: removing gnupg2-2.3.8-1.mga9.x86_64
>                                 
> #############################################
> Removed "/etc/systemd/user/default.target.wants/gpg-agent.service".

Downgrade the package make the steps
Update again and see the line  Removed "/etc/systemd/user/default.target.wants/gpg-agent.service"

CC: (none) => andrewsfarm
Whiteboard: (none) => MGA9-64-OK

Validating.

Keywords: (none) => validated_update
CC: (none) => sysadmin-bugs

An update for this issue has been pushed to the Mageia Updates repository.

https://advisories.mageia.org/MGAA-2024-0175.html

Resolution: (none) => FIXED
Status: NEW => RESOLVED