[systemd-devel] systemd-logind failing due to dbus error on org.freedesktop.systemd1

[Gustavo Sousa]

Hello, Michal.

After some trial an error I was able to find the culprit: the ypbind daemon.

- If I have it enabled, then it seems to me that, for some reason,
systemd can not take ownership of the dbus names (/me not sure if that
is the correct terminology).
- If I leave it disabled, then the problem disapears. Now I need to
find out why that is happening. The service file I got from my
distro's package is pasted at https://pastebin.com/vVEXLcb9.

With that, would I be right to say that this isn't a problem with
systemd itself?

Thank you!

Best regards,
Gustavo Sousa


On Wed, Jan 31, 2018 at 6:28 PM, Gustavo Sousa
<gustavo.jo.sousa at gmail.com> wrote:
> On Wed, Jan 31, 2018 at 3:52 PM, Michal Koutný <mkoutny at suse.com> wrote:
>>
>>
>> On 01/31/2018 03:55 PM, Gustavo Sousa wrote:
>>> Unfortunately, no. I didn't actually wait for the auto restart, I
>>> tried myself with 'systemctl restart systemd-logind'.
>> It seems like systemd-logind was thus properly connected to dbus and
>> there may be other issue.
>>
>> Could you please post logs preceding the snippet you sent previously?
>> (It seems something must went wrong at/before 8:38:51.)
> I looked at the logs preceding that, but they logged the same error
> (probably due to retries and me logging in?).
> Then I decided to look at the logs from the beginning and realized a
> different systemd error message (the snippet can be checked at
> https://pastebin.com/kcAVcUQz):
>
> Jan 30 14:18:09 systemd[1]: Failed to subscribe to NameOwnerChanged
> signal for 'org.freedesktop.login1': Connection timed out
>
> I also noticed that
>
>> Also since it
>> happened after reboot and systemd-logind restart didn't help, does it
>> mean it's reproducible or still present?
> Yes, it is still present. About it being reproducible, it is in my
> environment, but I'm not sure how easily that could be reproduced in a
> different environment.
>
>> Can you see 'org.freedesktop.systemd1' name owned in the `busctl` output?
> I'm not really familiar with dbus, but if by "owned" you mean that it
> does have a process associated to it, then I would say no. Here is a
> link for the output of `busctl`  https://pastebin.com/VqT38tya
>
>
>> Does `kill -SIGUSR1 1` help you?
> Nope. The issue persists (and the output of `busctl` keeps the same).
>
> Regards,
> Gustavo Sousa