Bug 725951 - systemd doesn't start dbus on cycles
systemd doesn't start dbus on cycles
Status: RESOLVED DUPLICATE of bug 661401
: 725656 (view as bug list)
Classification: openSUSE
Product: openSUSE 12.1
Classification: openSUSE
Component: GNOME
Factory
Other Other
: P1 - Urgent : Critical (vote)
: ---
Assigned To: Frederic Crozat
E-mail List
:
Depends on:
Blocks:
  Show dependency treegraph
 
Reported: 2011-10-22 11:46 UTC by Richard Brown
Modified: 2011-10-24 13:21 UTC (History)
2 users (show)

See Also:
Found By: ---
Services Priority:
Business Priority:
Blocker: ---
Marketing QA Status: ---
IT Deployment: ---
coolo: SHIP_STOPPER+


Attachments
dmesg output when xdm fails to start (65.92 KB, text/plain)
2011-10-22 15:01 UTC, Richard Brown
Details
dmesg for broken xdm/gdm startup at boot (53.80 KB, application/octet-stream)
2011-10-23 11:35 UTC, Harald Koenig
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Richard Brown 2011-10-22 11:46:51 UTC
Machine running openSUSE Factory + Gnome:Factory + Gnome:Apps for Factory
last zypper dup'd at approx 11:30 UTC on 22/10/2011

When booting, systemd fails to start xdm.service
Rebooting with systemd being verbose shows that the root cause is gdm failing to start with the following error

** (gdm:1043): WARNING **: Couldn't connect to system bus: Failed to connect to socket /var/run/dbus/system_bus_socket: No such file or directory
startproc: exit status of parent of /usr/sbin/gdm: 1

After the machine has booted, the file most certainly exists, and starting the service manually with systemctl start xdm.service works

So I'm guessing this is some sort of timing issue, but as I'm new to systemd I'm not sure where to look next, help!
Comment 1 Stephan Kulow 2011-10-22 14:20:52 UTC
please check dmesg if you can see any systemd cycles.
Comment 2 Stephan Kulow 2011-10-22 14:22:52 UTC
*** Bug 725656 has been marked as a duplicate of this bug. ***
Comment 3 Richard Brown 2011-10-22 15:01:23 UTC
Created attachment 458357 [details]
dmesg output when xdm fails to start

Attached is the dmesg output when xdm fails to start, doesn't appear to be very useful.
Comment 4 Stephan Kulow 2011-10-22 19:02:16 UTC
No, it's very useful. open-isci init script is problematic - I don't see how though 


[    9.335402] systemd[1]: Found ordering cycle on basic.target/start
[    9.335406] systemd[1]: Walked on cycle path to sockets.target/start
[    9.335409] systemd[1]: Walked on cycle path to dbus.socket/start
[    9.335411] systemd[1]: Walked on cycle path to sysinit.target/start
[    9.335413] systemd[1]: Walked on cycle path to open-iscsi.service/start
[    9.335415] systemd[1]: Walked on cycle path to network.target/start
[    9.335418] systemd[1]: Walked on cycle path to NetworkManager-wait-online.service/start
[    9.335420] systemd[1]: Walked on cycle path to NetworkManager.service/verify-active
[    9.335422] systemd[1]: Walked on cycle path to basic.target/start
[    9.335424] systemd[1]: Breaking ordering cycle by deleting job dbus.socket/start
Comment 5 Richard Brown 2011-10-22 19:13:25 UTC
I can confirm that uninstalling open-iscsi gets xdm starting up on reboot again..interesting. I don't see how its causing this problem however
Comment 6 Stephan Kulow 2011-10-22 19:30:10 UTC
I don't get it either, open-iscsi looks harmless.
Comment 7 Stephan Kulow 2011-10-22 19:46:00 UTC
possible the co-existance of boot.open-iscsi and open-iscsi confuses systemd. It does confuse me :)
Comment 8 Harald Koenig 2011-10-23 11:34:39 UTC
(In reply to comment #2)
> *** Bug 725656 has been marked as a duplicate of this bug. ***

indeed (full dmesg-systemd-1 attached)!
how to get rid of this (such) cycle(s) ?

$ dmesg | grep systemd 
[    5.882291] systemd[1]: systemd 37 running in system mode. (+PAM +LIBWRAP +AUDIT +SELINUX +SYSVINIT +LIBCRYPTSETUP; suse)
[    6.293614] systemd[1]: Set hostname to <linux-fudg.site>.
[    8.826441] systemd[1]: Found ordering cycle on basic.target/start
[    8.826450] systemd[1]: Walked on cycle path to sockets.target/start
[    8.826456] systemd[1]: Walked on cycle path to dbus.socket/start
[    8.826461] systemd[1]: Walked on cycle path to sysinit.target/start
[    8.826466] systemd[1]: Walked on cycle path to open-iscsi.service/start
[    8.826471] systemd[1]: Walked on cycle path to network.target/start
[    8.826476] systemd[1]: Walked on cycle path to NetworkManager-wait-online.service/start
[    8.826482] systemd[1]: Walked on cycle path to NetworkManager.service/verify-active
[    8.826487] systemd[1]: Walked on cycle path to basic.target/start
[    8.826492] systemd[1]: Breaking ordering cycle by deleting job dbus.socket/start
[    8.826507] systemd[1]: Found ordering cycle on basic.target/start
[    8.826512] systemd[1]: Walked on cycle path to sysinit.target/start
[    8.826516] systemd[1]: Walked on cycle path to open-iscsi.service/start
[    8.826521] systemd[1]: Walked on cycle path to network.target/start
[    8.826526] systemd[1]: Walked on cycle path to NetworkManager-wait-online.service/start
[    8.826532] systemd[1]: Walked on cycle path to NetworkManager.service/verify-active
[    8.826537] systemd[1]: Walked on cycle path to basic.target/start
[    8.826542] systemd[1]: Breaking ordering cycle by deleting job open-iscsi.service/start
[   22.031144] systemd-fsck[748]: /dev/sda1: sauber, 265/40000 Dateien, 51194/159744 Bl\xffffffc3\xffffffb6\xffffffb6cke

the last line of systemd-fsck with german msg doesn't fit in the all-english syslog msgs,
and what happend to the german umlaut-o (UTF8 0xc3 0xb6) ? and why does 0xb6 show up twice ???
(-> bug #725993)


back to the systemd cycle: I tried "rpm -e open-iscsi", and now it looks "somewhat" better:

# dmesg | grep systemd
[    5.713276] systemd[1]: systemd 37 running in system mode. (+PAM +LIBWRAP +AUDIT +SELINUX +SYSVINIT +LIBCRYPTSETUP; suse)
[    6.076103] systemd[1]: Set hostname to <linux-fudg.site>.
[   26.867730] systemd-fsck[759]: /dev/sda1: sauber, 265/40000 Dateien, 51194/159744 Bl\xffffffc3\xffffffb6\xffffffb6cke

(X11 still does not show up but now the Xserver crashes -- different report #725996 ...)
Comment 9 Harald Koenig 2011-10-23 11:35:41 UTC
Created attachment 458383 [details]
dmesg for broken xdm/gdm startup at boot
Comment 10 Frederic Crozat 2011-10-24 12:23:26 UTC
systemd has a hard time to figure dependencies when two initscripts have the same name and one is a "boot" initscript. 

I'm looking at how to solve the issue.
Comment 11 Frederic Crozat 2011-10-24 13:21:13 UTC
was already filled some time ago, let's handle it there.

*** This bug has been marked as a duplicate of bug 661401 ***