Devuan bug report logs - #301
slim logins don't get registered in utmp

Reply or subscribe to this bug.

Toggle useless messages

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

Package: slim
Version: 1.3.6-5+devuan6

slim doesn't seem to update /var/run/utmp and /var/log/wtmp anymore?
...........
also previous versions of slim: I'm still getting the 'w' report:
------
 08:06:53 up 0 min,  0 users,  load average: 0.29, 0.08, 0.02
USER     TTY      FROM             LOGIN@   IDLE   JCPU   PCPU WHAT
------

Likewise, 'last' tells about reboot and tty1 login(s) but not my 'slim'
logins.

I getting the same with the
previous slim distributions as well; I returned to 1.3.6-5.1+devuan2
just to verify, and I've now also gone "back" to 1.3.6-5+devuan4, which
'apt-cache policy slim' claims is the jessie version; the 'w' report
stays the same (also after reboots).

So, apparently 'slim' since long ago believes something else is
responsible for registring the session(s), but that something else
doesn't agree. Maybe that's down to 'consolekit', then?
.........
[addition by wirelessduck via dng@lists.dyne.org]
I’m running consolekit with lightdm on a regular ascii.

I just installed slim alongside lightdm and set it as default. Rebooted
and logged in via slim.

I get nothing listed when running 'w' and 'who'. Checking output of
'last' shows the reboot time correctly but my slim login doesn’t appear
at all.

I switched back to lightdm and I see my user appearing in the output of
both 'w' and 'who'.  The 'last' output also shows my lightdm login from
tty7.

I should note that this login account comes from winbind, but I don’t
think that should be affecting this at all?

—Tom

..............
Further on this:

I've also got a dist-upgraded beowulf with a no-install-recommends
installation of task-xfce-desktop, and this one is fine wrt 'w'/'last'
report for slim using consolekt, and still fine when I downgraded
consolekit to the ascii version. The newest 'slim' is fine btw.

A little bit of digging tells me that my laptop has XDG_SESSION_COOKIE
set which my beowulf VM doesn't. That difference might be enough to
steer things differently for /etc/X11/Xsession.d/90consolekit, which (by
guesswork) has a significant decision point about the session startup.
Basically, X startup only runs /usr/bin/ck-launch-session if the
variable is unset (at that time).

Someone who knows stuff might kindly pull me right, but now I'm chasing
the reason why may latop gets XDG_SESSION_COOKIE set (while my beowulf
VM doesn't)...

Ralph.

Message #10 received at 301@bugs.devuan.org (full text, mbox, reply):

On Thu, Mar 07, 2019 at 08:55:43PM +1100, Ralph Ronnquist wrote:
> I've also got a dist-upgraded beowulf with a no-install-recommends
> installation of task-xfce-desktop, and this one is fine wrt 'w'/'last'
> report for slim using consolekt, and still fine when I downgraded
> consolekit to the ascii version. The newest 'slim' is fine btw.
> 
> A little bit of digging tells me that my laptop has XDG_SESSION_COOKIE
> set which my beowulf VM doesn't. That difference might be enough to
> steer things differently for /etc/X11/Xsession.d/90consolekit, which (by
> guesswork) has a significant decision point about the session startup.
> Basically, X startup only runs /usr/bin/ck-launch-session if the
> variable is unset (at that time).
> 
> Someone who knows stuff might kindly pull me right, but now I'm chasing
> the reason why may latop gets XDG_SESSION_COOKIE set (while my beowulf
> VM doesn't)...

XDG_SESSION_COOKIE is set by consolekit. elogind doesn't set it, so that makes
me think your beowulf VM which is working OK is using elogind rather than
consolekit. Can you check?

Thanks

Mark

Message #15 received at 301@bugs.devuan.org (full text, mbox, reply):

Ralph,

Could you try adding this configuration to /etc/slim.conf please?

# sessreg. You can use the %user variable
sessionstop_cmd exec sessreg -d -l $DISPLAY %user
sessionstart_cmd exec sessreg -a -l $DISPLAY %user

It fixes slim utmp/wtmp/lastlog registration for me although the process is
identified as ?xdm?. Trying to fix that....

Mark

Message #20 received at 301@bugs.devuan.org (full text, mbox, reply):

On Fri, Mar 08, 2019 at 05:27:34PM +0000, Mark Hindley wrote:
> Ralph,
> 
> Could you try adding this configuration to /etc/slim.conf please?
> 
> # sessreg. You can use the %user variable
> sessionstop_cmd exec sessreg -d -l $DISPLAY %user
> sessionstart_cmd exec sessreg -a -l $DISPLAY %user
> 
> It fixes slim utmp/wtmp/lastlog registration for me although the process is
> identified as ?xdm?. Trying to fix that....

Ralph,

Just a follow up on this.

Although it is ugly, I think the ?xdm? in the IDLE column out from w is
technically correct: w is unable to determine the correct idle time for X. You
need xprintidle or the like for that.

Other than that, sessreg correctly registers sessions in utmp and lastlog for
me.

Can you confirm it works for you?

Mark

Message #25 received at 301@bugs.devuan.org (full text, mbox, reply):

[Message part 1 (text/plain, inline)]

tested again using slim/experimental .  devuan testing/ceres amd64.

# w
 18:18:10 up  8:39,  1 user,  load average: 0.75, 0.66, 0.73
USER     TTY      FROM             LOGIN@   IDLE   JCPU   PCPU WHAT
user       :0.0     :0.0             09:38   ?xdm?   2:53m  2.27s
x-session-manager

# last            
user       :0.0         :0.0             Thu Jun 13 09:38   still logged in
reboot   system boot  4.19.0-5-amd64   Thu Jun 13 09:38   still running

# lastlog -u user
Username         Port     From             Latest
user               :0.0                      Thu Jun 13 09:38:58 +0300 2019

# apt policy slim            

slim:
  Installed: 1.3.6-5.1+devuan3


thanks,

dimitris.

[signature.asc (application/pgp-signature, attachment)]

Message #30 received at 301-done@bugs.devuan.org (full text, mbox, reply):

Fixed by calling usr/bin/sessreg.

See commit https://git.devuan.org/devuan-packages/slim/commit/d90325a6467859fe6204da89a1921e75b0b59666

Send a report that this bug log contains spam.