Debian Bug report logs - #862258
create mlmmj user

Toggle useless messages

Message #5 received at submit@bugs.debian.org (full text, mbox, reply):

From: Geert Stappers <stappers@stappers.nl>

To: submit@bugs.debian.org

Subject: create mlmmj user

Date: Wed, 10 May 2017 13:08:54 +0200

Package: mlmmj
Version: 1.2.19.0-1
Severity: wishlist

Hello Maintainers of mlmmj,


Both the Postfix and the Exim   README of mlmmj advise to create
a dedicate user for mlmmj.

It would be nice if the Debian package of mlmmj would do this.

 
Groeten
Geert Stappers
-- 
Leven en laten leven

Message #10 received at 862258@bugs.debian.org (full text, mbox, reply):

From: Geert Stappers <stappers@debian.org>

To: 862258@bugs.debian.org

Subject: [PATCH] add/remove user mlmmj in maintainer scripts

Date: Tue, 23 May 2017 20:57:02 +0200

Closes: #862258

debian/mlmmj.postinst and debian/mlmmj.postrm rewrite based
upon maintainer scripts from the sbuild package.

Plus some minor changes.
---
 debian/control                  |    4 ++-
 debian/cron.d                   |    2 +-
 debian/mlmmj.postinst           |   73 ++++++++++++++++++++++++++++++++++++++-
 debian/mlmmj.postrm             |   57 ++++++++++++++++++------------
 debian/patches/mlmmj_user_added |   36 +++++++++++++++++++
 debian/patches/series           |    1 +
 6 files changed, 148 insertions(+), 25 deletions(-)
 create mode 100644 debian/patches/mlmmj_user_added

diff --git a/debian/control b/debian/control
index 198b7de..7fc48c7 100644
--- a/debian/control
+++ b/debian/control
@@ -12,7 +12,9 @@ Homepage: http://mlmmj.org/
 
 Package: mlmmj
 Architecture: any
-Depends: ${shlibs:Depends}, ${misc:Depends}, postfix | exim4 | mail-transport-agent
+Depends: ${shlibs:Depends}, ${misc:Depends}
+  , postfix | exim4 | mail-transport-agent
+  , adduser
 Suggests: mlmmj-php-web, mlmmj-php-web-admin
 Description: mail server independent mailing list manager
  A mailing list manager with roughly the same functionality as ezmlm.
diff --git a/debian/cron.d b/debian/cron.d
index 97d2c8c..2f34265 100644
--- a/debian/cron.d
+++ b/debian/cron.d
@@ -1 +1 @@
-0 */2 * * * root /usr/bin/test -x /usr/bin/mlmmj-maintd && /usr/bin/mlmmj-maintd -F -d /var/spool/mlmmj
+0 */2 * * * mlmmj /usr/bin/test -x /usr/bin/mlmmj-maintd && /usr/bin/mlmmj-maintd -F -d /var/spool/mlmmj
diff --git a/debian/mlmmj.postinst b/debian/mlmmj.postinst
index 9d168c3..8c2ec9d 100644
--- a/debian/mlmmj.postinst
+++ b/debian/mlmmj.postinst
@@ -1,4 +1,7 @@
 #!/bin/sh
+# postinst script for mlmmj
+#
+# see: dh_installdeb(1)
 
 set -e
 
@@ -9,7 +12,75 @@ else
         echo "No debconf installed"
 fi
 
-mkdir -p /var/spool/mlmmj
+# summary of how this script can be called:
+#        * <postinst> `configure' <most-recently-configured-version>
+#        * <old-postinst> `abort-upgrade' <new version>
+#        * <conflictor's-postinst> `abort-remove' `in-favour' <package>
+#          <new-version>
+#        * <deconfigured's-postinst> `abort-deconfigure' `in-favour'
+#          <failed-install-package> <version> `removing'
+#          <conflicting-package> <version>
+# for details, see /usr/share/doc/packaging-manual/
+#
+# quoting from the policy:
+#     Any necessary prompting should almost always be confined to the
+#     post-installation script, and should be protected with a conditional
+#     so that unnecessary prompting doesn't happen if a package's
+#     installation fails and the `postinst' is called with `abort-upgrade',
+#     `abort-remove' or `abort-deconfigure'.
+
+case "$1" in
+    configure)
+	# If mlmmj is a user group, migrate to a system group.  Save
+	# a list of group members, then add them back to the new group
+	# once created.
+	SGID="$(grep ^mlmmj: /etc/group | sed -r -e 's/([^:]+):([^:]+):([^:]+):.*/\3/')"
+	if [ -n "$SGID" ] && [ "$SGID" -ge "1000" ]; then
+	    SUSR="$(grep ^mlmmj: /etc/group | sed -r -e 's/([^:]+):([^:]+):([^:]+):(.*)$/\4/' -e 's/,/ /g')"
+	    delgroup --quiet mlmmj
+	fi
+
+        if ! getent group mlmmj > /dev/null; then
+            addgroup --system --quiet mlmmj
+        fi
+
+	# Restore group membership to users in old user group (if any)
+	for user in $SUSR; do
+	    adduser $user mlmmj
+	done
+
+	# Create dedicated mlmmj user
+        if ! getent passwd mlmmj > /dev/null; then
+            adduser --system --quiet \
+		--home /var/spool/mlmmj --no-create-home \
+		--shell /bin/false --ingroup mlmmj \
+                --gecos "Mailing List Management Made Joyful" \
+                mlmmj
+        fi
+	# Unlock account in case it was locked from previous purge.
+	usermod -U -e '' mlmmj || true
+
+	mkdir -p /var/spool/mlmmj
+        chown -R mlmmj:mlmmj /var/spool/mlmmj
+        chmod 02770 /var/spool/mlmmj
+
+        ;;
+
+    abort-upgrade|abort-remove|abort-deconfigure)
+
+        ;;
+
+    *)
+
+        echo "postinst called with unknown argument \`$1'" >&2
+        exit 0
+
+	;;
+esac
+
+# dh_installdeb will replace this with shell code automatically
+# generated by other debhelper scripts.
 
 #DEBHELPER#
 
+exit 0
diff --git a/debian/mlmmj.postrm b/debian/mlmmj.postrm
index cfdcd4e..b2f2b76 100644
--- a/debian/mlmmj.postrm
+++ b/debian/mlmmj.postrm
@@ -1,34 +1,47 @@
 #!/bin/sh
+# postrm script for mlmmj
+#
+# see: dh_installdeb(1)
+
 set -e
 
+# summary of how this script can be called:
+#        * <postrm> `remove'
+#        * <postrm> `purge'
+#        * <old-postrm> `upgrade' <new-version>
+#        * <new-postrm> `failed-upgrade' <old-version>
+#        * <new-postrm> `abort-install'
+#        * <new-postrm> `abort-install' <old-version>
+#        * <new-postrm> `abort-upgrade' <old-version>
+#        * <disappearer's-postrm> `disappear' <r>overwrit>r> <new-version>
+# for details, see /usr/share/doc/packaging-manual/
+
 case "$1" in
-    purge)
-        if [ -e /usr/share/debconf/confmodule ]
-        then
-            . /usr/share/debconf/confmodule
-            db_get mlmmj/remove-on-purge || true
-            if [ "$RET" = "true" ]
-            then
-                echo "Removing /var/spool/mlmmj and /etc/mlmmj/lists."
-                rm -rf /var/spool/mlmmj /etc/mlmmj/lists > /dev/null
-            fi
-            db_purge || true
-        fi
-        if grep -qs '/usr/bin/mlmmj' /etc/aliases
-        then
-            echo "Warning /etc/aliases appears to contain references to mlmmj." >&2
-        fi
-    ;;
-
-    remove|upgrade|failed-upgrade|abort-install|abort-upgrade|disappear)
-    ;;
+
+    remove|upgrade|failed-upgrade|abort-install|abort-upgrade)
+
+        ;;
+
+    purge|disappear)
+
+	# Lock mlmmj account.
+	usermod -L -e 1 mlmmj || true
+
+	# not removing anything else
+
+        ;;
 
     *)
+
         echo "postrm called with unknown argument \`$1'" >&2
-        exit 1
-    ;;
+        exit 0
+
+	;;
 esac
 
+# dh_installdeb will replace this with shell code automatically
+# generated by other debhelper scripts.
+
 #DEBHELPER#
 
 exit 0
diff --git a/debian/patches/mlmmj_user_added b/debian/patches/mlmmj_user_added
new file mode 100644
index 0000000..92f148a
--- /dev/null
+++ b/debian/patches/mlmmj_user_added
@@ -0,0 +1,36 @@
+Description: mlmmj user is created
+Author: Geert Stappers <stappers@debian.org>
+Bug-Debian: https://bugs.debian.org/862258
+Origin: vendor
+Forwarded: no
+Last-Update: 2017-05-15
+
+
+Index: mlmmj/README.exim4
+===================================================================
+--- mlmmj.orig/README.exim4
++++ mlmmj/README.exim4
+@@ -9,6 +9,7 @@ of this can be found on http://plonk.de/
+ 
+ Notes:
+ - We assume that you have a user and group called mlmmj to use with mlmmj
++  (FYI: it is done by the .deb package)
+ - The exim user needs rx access rights to mlmmj's spool directory. (If you
+   don't want that, see below.) The easiest way is 
+   "chmod 755 /path/to/mlmmj/spool", if it's ok that local users can see which
+Index: mlmmj/README.postfix
+===================================================================
+--- mlmmj.orig/README.postfix
++++ mlmmj/README.postfix
+@@ -23,9 +23,11 @@ Postfix virtual domains (so you can host
+     makes sense to make this a 'system' user, with no password and no shell
+     (/usr/false for the shell), and for its home directory to be
+     /var/spool/mlmmj (or wherever you want to put your Mlmmj spool directory).
++ This is done by the packaging system.
+ 
+  2) Create your Mlmmj spool directory (we'll assume it's /var/spool/mlmmj)
+     and change its owner to the 'mlmmj' user.
++ This is done by the packaging system.
+ 
+  3) Add an 'mlmmj' transport which uses the pipe(8) delivery agent to execute
+     mlmmj-receive as the mlmmj user by adding something like the following to
diff --git a/debian/patches/series b/debian/patches/series
index c6f710e..3da5b8f 100644
--- a/debian/patches/series
+++ b/debian/patches/series
@@ -3,3 +3,4 @@
 fixes-path-to-templates-folder-in-php-web-admin
 06_fix-php-web-for-php7.diff
 07_fix-crontab-suggestions.diff
+mlmmj_user_added
-- 
1.7.10.4

-- 
Groeten
Geert Stappers
-- 
Leven en laten leven

Message #17 received at 862258-done@bugs.debian.org (full text, mbox, reply):

From: Thomas Goirand <zigo@debian.org>

To: Geert Stappers <stappers@debian.org>, 862258-done@bugs.debian.org

Subject: Re: [Pkg-mlmmj-devel] Bug#862258: [PATCH] add/remove user mlmmj in maintainer scripts

Date: Wed, 24 May 2017 16:00:37 +0200

Hi,

I have seen your patches, however, I don't think it is really necessary
to add a new system user for mlmmj. There are many ways to use MLMMJ,
one would be without using any system user at all, which is what one
would use to do virtual hosting, and which is what I do. In such a
configuration, the username would not be mlmmj. Therefore, it's best to
just leave the choice to the user and let him do it by hand. I believe
here, you're maybe too much influenced by bad practice with less good
list manager, like mailman, maybe?

By the way, there's a few things that would be prefectible with your
scripts. I am not fancy of the switch/case type of postinst / postrm. I
very much prefer something like this:

if [ "$1" = "configure" ] || [ "$1" = "reconfigure" ] ; then
    [ ... ]
fi

Also, using grep is suboptimal, one should prefer using the getent shell
tool, which is made for the purpose.

What I'm convinced as well, is that never, a package should attempt to
remove its user from the system once it has been added, because of the
UUID reuse security issue.

I hope this doesn't disappoint you.

Cheers,

Thomas Goirand (zigo)

Message #22 received at 862258@bugs.debian.org (full text, mbox, reply):

From: Geert Stappers <stappers@debian.org>

To: Thomas Goirand <zigo@debian.org>, 862258@bugs.debian.org

Subject: add an user for mlmmj

Date: Thu, 25 May 2017 10:59:21 +0200

On Wed, May 24, 2017 at 04:00:37PM +0200, Thomas Goirand wrote:
> Hi,
> 
> I have seen your patches, however, I don't think it is really necessary
> to add a new system user for mlmmj. There are many ways to use MLMMJ,
> one would be without using any system user at all, which is what one
> would use to do virtual hosting, and which is what I do. In such a
> configuration, the username would not be mlmmj. Therefore, it's best to
> just leave the choice to the user and let him do it by hand.

It is the "add an user _by hand_" that I want to avoid.

> I believe here, you're maybe too much influenced by bad practice with
> less good list manager, like mailman, maybe?

Both http://mlmmj.org/docs/readme-exim4/ and http://mlmmj.org/docs/readme-postfix/
say to create a mlmmj user.


What would an acceptable place for inclusion of 


            adduser --system --quiet \
                --home /var/spool/mlmmj --no-create-home \
                --shell /bin/false --ingroup mlmmj \
                --gecos "Mailing List Management Made Joyful" \
                mlmmj


?


Groeten
Geert Stappers
-- 
Leven en laten leven

Message #27 received at 862258@bugs.debian.org (full text, mbox, reply):

From: Thomas Goirand <zigo@debian.org>

To: Geert Stappers <stappers@debian.org>, 862258@bugs.debian.org

Subject: Re: [Pkg-mlmmj-devel] Bug#862258: add an user for mlmmj

Date: Thu, 25 May 2017 13:43:01 +0200

On 05/25/2017 10:59 AM, Geert Stappers wrote:
> On Wed, May 24, 2017 at 04:00:37PM +0200, Thomas Goirand wrote:
>> Hi,
>>
>> I have seen your patches, however, I don't think it is really necessary
>> to add a new system user for mlmmj. There are many ways to use MLMMJ,
>> one would be without using any system user at all, which is what one
>> would use to do virtual hosting, and which is what I do. In such a
>> configuration, the username would not be mlmmj. Therefore, it's best to
>> just leave the choice to the user and let him do it by hand.
> 
> It is the "add an user _by hand_" that I want to avoid.

There's no way to avoid it, since this is a very specific way to use
MLMMJ (ie: using a single mlmmj system account isn't the way when doing
virtual hosting).

>> I believe here, you're maybe too much influenced by bad practice with
>> less good list manager, like mailman, maybe?
> 
> Both http://mlmmj.org/docs/readme-exim4/ and http://mlmmj.org/docs/readme-postfix/
> say to create a mlmmj user.

This is *one way* to do things, yes.

> What would an acceptable place for inclusion of 
> 
> 
>             adduser --system --quiet \
>                 --home /var/spool/mlmmj --no-create-home \
>                 --shell /bin/false --ingroup mlmmj \
>                 --gecos "Mailing List Management Made Joyful" \
>                 mlmmj

I don't think so.

Cheers,

Thomas Goirand (zigo)

Message #32 received at 862258@bugs.debian.org (full text, mbox, reply):

From: Geert Stappers <stappers@debian.org>

To: Thomas Goirand <zigo@debian.org>, 862258@bugs.debian.org

Subject: Re: [Pkg-mlmmj-devel] Bug#862258: add an user for mlmmj

Date: Sat, 27 May 2017 01:48:16 +0200

On Thu, May 25, 2017 at 01:43:01PM +0200, Thomas Goirand wrote:
> On 05/25/2017 10:59 AM, Geert Stappers wrote:
> > On Wed, May 24, 2017 at 04:00:37PM +0200, Thomas Goirand wrote:
> >> just leave the choice to the user and let him do it by hand.
> > 
> > It is the "add an user _by hand_" that I want to avoid.
> 
> There's no way to avoid it, since this is a very specific way to use
> MLMMJ (ie: using a single mlmmj system account isn't the way when doing
> virtual hosting).
> 
> >> I believe here, you're maybe too much influenced by bad practice with
> >> less good list manager, like mailman, maybe?
> > 
> > Both http://mlmmj.org/docs/readme-exim4/ and http://mlmmj.org/docs/readme-postfix/
> > say to create a mlmmj user.
> 
> This is *one way* to do things, yes.

Those who want to use that mlmmj user will benefit from the "postinst adduser".
 

> > What would an acceptable place for inclusion of 
> > 
> >             adduser --system --quiet \
> >                 --home /var/spool/mlmmj --no-create-home \
> >                 --shell /bin/false --ingroup mlmmj \
> >                 --gecos "Mailing List Management Made Joyful" \
> >                 mlmmj
> 
> I don't think so.
> 


Thing I miss in this discussion is how a mlmmj user is harming other workflows.

Could the virtual hosting workflow / setup be elaborated?



Groeten
Geert Stappers
-- 
Leven en laten leven

Message #37 received at 862258@bugs.debian.org (full text, mbox, reply):

From: Thomas Goirand <zigo@debian.org>

To: Geert Stappers <stappers@debian.org>, 862258@bugs.debian.org

Subject: Re: [Pkg-mlmmj-devel] Bug#862258: Bug#862258: add an user for mlmmj

Date: Sat, 27 May 2017 21:49:49 +0200

On 05/27/2017 01:48 AM, Geert Stappers wrote:
> On Thu, May 25, 2017 at 01:43:01PM +0200, Thomas Goirand wrote:
>> On 05/25/2017 10:59 AM, Geert Stappers wrote:
>>> On Wed, May 24, 2017 at 04:00:37PM +0200, Thomas Goirand wrote:
>>>> just leave the choice to the user and let him do it by hand.
>>>
>>> It is the "add an user _by hand_" that I want to avoid.
>>
>> There's no way to avoid it, since this is a very specific way to use
>> MLMMJ (ie: using a single mlmmj system account isn't the way when doing
>> virtual hosting).
>>
>>>> I believe here, you're maybe too much influenced by bad practice with
>>>> less good list manager, like mailman, maybe?
>>>
>>> Both http://mlmmj.org/docs/readme-exim4/ and http://mlmmj.org/docs/readme-postfix/
>>> say to create a mlmmj user.
>>
>> This is *one way* to do things, yes.
> 
> Those who want to use that mlmmj user will benefit from the "postinst adduser".

But those who don't will have a useless user setup *every time* the
package is upgraded.

Cheers,

Thomas Goirand (zigo)

Message #42 received at 862258@bugs.debian.org (full text, mbox, reply):

From: Geert Stappers <stappers@debian.org>

To: Thomas Goirand <zigo@debian.org>, 862258@bugs.debian.org

Subject: Re: add an user for mlmmj

Date: Mon, 29 May 2017 13:03:38 +0200

On Sat, May 27, 2017 at 09:49:49PM +0200, Thomas Goirand wrote:
> On 05/27/2017 01:48 AM, Geert Stappers wrote:
> >> On 05/25/2017 10:59 AM, Geert Stappers wrote:
> >>> Both http://mlmmj.org/docs/readme-exim4/ and http://mlmmj.org/docs/readme-postfix/
> >>> say to create a mlmmj user.
> > 
> > Those who want to use that mlmmj user will benefit from the "postinst adduser".
> 
> But those who don't will have a useless user setup *every time* the
> package is upgraded.

Code from the patch, note the 'if' statement

+	# Create dedicated mlmmj user
+        if ! getent passwd mlmmj > /dev/null; then
+            adduser --system --quiet \
+		--home /var/spool/mlmmj --no-create-home \
+		--shell /bin/false --ingroup mlmmj \
+                --gecos "Mailing List Management Made Joyful" \
+                mlmmj
+        fi


That 'if' statement makes
> But those who don't will have a useless user setup *every time* the
> package is upgraded.
into
} But those who don't will have a useless user setup


Right now we are talking about entries in /etc/passwd and /etc/group
that _might not_ be used.

Could we go back to /etc/passwd /etc/group entries
that _are_ being used?



Groeten
Geert Stappers
-- 
Leven en laten leven

Message #47 received at 862258@bugs.debian.org (full text, mbox, reply):

From: Geert Stappers <stappers@stappers.nl>

To: 862258@bugs.debian.org

Subject: Re: add an user for mlmmj

Date: Tue, 30 May 2017 22:47:11 +0200

Control: reopen -1
Stop

Ticket reopened so the BTS accepts further updates.

Message #54 received at 862258-done@bugs.debian.org (full text, mbox, reply):

From: Thomas Goirand <thomas@goirand.fr>

To: 862258-done@bugs.debian.org

Subject: Please do not re-open

Date: Mon, 5 Jun 2017 23:42:51 +0200

Hi,

Please do not re-open this bug. This "feature" of the package isn't up
for discussion anymore. If you really do want to discuss it, feel free
to open a bug with the tech ctte. I don't have time for more BTS
ping-pong, and I already explained why I believe it would be harmful to
automatically add an MLMMJ system account and set the cron job to use
such an account.

Cheers,

Thomas Goirand (zigo)

Message #59 received at 862258@bugs.debian.org (full text, mbox, reply):

From: Geert Stappers <stappers@stappers.nl>

To: 862258@bugs.debian.org

Subject: so far

Date: Mon, 12 Jun 2017 00:01:35 +0200

So far is this bugreport missing how the existence of a mlmmj user
could hinder any mlmmj workflow.

There is being told "not needed".  And "do it by hand".
That are no good reasons to block something that upstream advices.


Going to CTTE is indeed an option, for later.

Please reconsider the creation of an user during install.


Cheers
Geert Stappers
Hoping that the BTS accepts this update to this premature closed bugreport.

Message #64 received at 862258@bugs.debian.org (full text, mbox, reply):

From: Thomas Goirand <zigo@debian.org>

To: Geert Stappers <stappers@stappers.nl>, 862258@bugs.debian.org

Subject: Re: [Pkg-mlmmj-devel] Bug#862258: so far

Date: Wed, 14 Jun 2017 22:48:34 +0200

On 06/12/2017 12:01 AM, Geert Stappers wrote:
> 
> So far is this bugreport missing how the existence of a mlmmj user
> could hinder any mlmmj workflow.
> 
> There is being told "not needed".  And "do it by hand".
> That are no good reasons to block something that upstream advices.
> 
> 
> Going to CTTE is indeed an option, for later.
> 
> Please reconsider the creation of an user during install.
> 
> 
> Cheers
> Geert Stappers
> Hoping that the BTS accepts this update to this premature closed bugreport.

What you're advising is to use an MLMMJ user, and later on using that
user for the cron job. That's at least what your patch shows. In such
configuration, it would break any user that is using virtual hosting
with one UUID per user (and that's only one type of setup). If we
wouldn't be using the MLMMJ user for the cron job, then your patch
becomes simply useless (why creating a user if we're not using it?).

Last, your proposed patch is *unconditionally* creating the user,
meaning that those who don't want it will have it forced into their
system on every upgrade. I don't want to force such a thing to someone
who doesn't want it.

On your side, you don't have any point of argumentation except
"something that upstream advices", and the fact that you want it for
your own usage. There's no point of argumentation about why it's so
important to have that system user being setup *by default* (ie: of
course, it's very easy for anyone to simply run adduser on the shell, or
automate the task with a simple script). On top of this, you just insist
repeating over and over the same thing (ie: empty argumentation). This
is becoming annoying and pointless.

I really would like all of this noise to stop. Please don't insist the
way you're doing: it's the wrong way. A more constructive way would have
for example proposing a patch with a debconf question, or explaining
your use case, and why it's not possible to run adduser by hand or with
a script.

Cheers,

Thomas Goirand (zigo)

Send a report that this bug log contains spam.