[SOLVED] mga3 : can not reboot after ordinary update

[andre999]

Note first that this was not an upgrade to mga4.
Just an ordinary upgrade using rpmdrake, also adding a few new applications.
Removing the new applications doesn't fix the problem.
I can only boot from the hard disk into rescue mode.
(When booting with the gentoo-based Sysrescuecd to allow me internet access and linux gui, all hw seems to work normally.)

I suspect it is related to some systemd related config change, but I'm not familiar enough with systemd to find the problem.

(line#, time) = selected lines from journalctl output of failed boot. (attachment has full details)

Code: Select all

_196, 04:08:03 = kernel: tsc: Marking TSC unstable due to check_tsc_sync_source failed
_550, 04:08:03 = kernel: ata1: softreset failed (device not ready)
_552, 04:08:03 = kernel: ata3: softreset failed (device not ready)
_786, 04:08:03 = systemd[1]: Cannot add dependency job for unit systemd-timedated-ntp.target, ignoring: Unit systemd-timedated-ntp.target failed to load: No such file or directory. See system logs and 'systemctl status systemd-timedated-ntp.target' for details.

_881, 04:08:04 = kernel: sp5100_tco: failed to find MMIO address, giving up.

1092, 04:08:29 = NetworkManager[677]: <warn> failed to allocate link cache: (-10) Operation not supported
1103, 04:08:29 = NetworkManager[677]: <warn> failed to allocate link cache: (-10) Operation not supported

1110, 04:08:29 = cpufreqd[670]: [FAILED]
1111, 04:08:29 = systemd[1]: cpufreqd.service: control process exited, code=exited status=127
1112, 04:08:29 = systemd[1]: Failed to start LSB: CPU frequency scaling daemon.
1113, 04:08:29 = systemd[1]: Unit cpufreqd.service entered failed state

1040, 04:08:28 localhost.localdomain avahi-daemon[686]: WARNING: No NSS support for mDNS detected, consider installing nss-mdns!
1093, 04:08:29 = NetworkManager[677]: ifcfg-rh:     error: Missing SSID
1106, 04:08:29 = modprobe[671]: modprobe: ERROR: missing parameters. See -h.
1117, 04:08:29 = upowerd[685]: (upowerd:685): UPower-Linux-WARNING **: energy 40.381200 bigger than full 40.370400

1211, 04:08:29 = dbus[687]: [system] Rejected send message, 1 matched rules; type="method_call", sender=":1.13" (uid=0 pid=693 comm="/usr/sbin/gdm-binary -nodaemon ") interface="org.freedesktop.DBus.Properties" member="GetAll" error name="(unset)" requested_reply="0" destination=":1.15" (uid=0 pid=870 comm="/usr/libexec/gdm-simple-slave --display-id /org/gn")

1223, 04:08:29 = dbus-daemon[687]: dbus[687]: [system] Rejected send message, 1 matched rules; type="method_call", sender=":1.13" (uid=0 pid=693 comm="/usr/sbin/gdm-binary -nodaemon ") interface="org.freedesktop.DBus.Properties" member="GetAll" error name="(unset)" requested_reply="0" destination=":1.15" (uid=0 pid=870 comm="/usr/libexec/gdm-simple-slave --display-id /org/gn")

1247, 04:08:30 = gdm-simple-slave[870]: WARNING: Failed to give slave programs access to the display. Trying to proceed.

1273, 04:08:31 = accounts-daemon[1173]: (accounts-daemon:1173): GLib-GObject-WARNING **: g_object_notify: object class `User' has no property named `x-session'

1363, 04:08:36 localhost.localdomain gnome-session[1229]: WARNING: App 'gnome-shell.desktop' respawning too quickly
1369, 04:08:40 = ntpdate[1517]: Can't find host 0.0.ca.pool.ntp.org: Name or service not known (-2)
1370, 04:08:41 = ntpdate[1517]: Can't find host 1.0.ca.pool.ntp.org: Name or service not known (-2)
1371, 04:08:41 = ntpdate[1517]: Can't find host 2.0.ca.pool.ntp.org: Name or service not known (-2)
1372, 04:08:41 = ntpdate[1517]: no servers can be used, exiting

1374, 04:08:57 = systemd-journal[356]: Forwarding to syslog missed 54 messages.
1375, 04:08:58 = systemd-logind[683]: System is rebooting.


(The file has no line numbers, and "localhost.localdomain" in place of "=".)

Any hints would be very much appreciated

[doktor5000]

If you wanted an achievement for the ratio of most error messages in one log excerpt, you got it
Everything seems broken, there's not one normal message in there from a short overview.

Well, two major things I can spot:
- network seems down
- dbus seems broken, so GDM and in turn Gnome will not start

Some more things from your attached log, randonly chosen:

Code: Select all

Feb 05 04:08:29 localhost.localdomain modprobe[671]: modprobe: ERROR: missing parameters. See -h.
Feb 05 04:08:29 localhost.localdomain cpufreqd[670]: Starting cpufreqd : /bin/bash: cpufreqd : commande introuvable
Feb 05 04:08:30 localhost.localdomain gdm-simple-slave[870]: WARNING: Failed to give slave programs access to the display. Trying to proceed.

Then a reboot in between? huh? What did you do there?

Code: Select all

Feb 05 04:08:34 localhost.localdomain pulseaudio[1390]: [pulseaudio] authkey.c: Failed to open cookie file '/var/lib/gdm/.config/pulse/cookie': Aucun fichier ou dossier de ce type
-- Reboot --

feb 04

-- Reboot --
Feb 05 04:08:34 localhost.localdomain dbus[687]: [system] Activating via systemd: service name='org.freedesktop.RealtimeKit1' unit='rtkit-daemon.service'

You have three bets:
Change to another display manager and try if that helps.
Use a Mageia 3 DVD to do an "Update" via the installer, maybe it helps.
My recommendation would be to reinstall and keep your /home.

[andre999]

Thanks for your feedback, Florian.

As for the number of errors, I have no idea what caused it. The only thing that risked to cause a problem was the update that I had done a few days before rebooting. And I had only rebooted because I was going to do a backup. No signs of a problem.

I was hoping for some quick fix for a minor error, but since you have a lot more experience looking at systemd logs, I'll trust your judgement.
As for the split log, with "reboot" in between, I think the log was split because the log messages were in a temporary location, and on failure they weren't posted in the right place. It occurred exactly like that in 3 different log files.
For the network error messages, Colin already told me that these messages aren't a problem, and there is already a bug opened about that. (Colin has commented on that bug.)

I was already thinking about trying an update from the mga3 dvd, so I'll try that first.
(Luckily it is handy since I always install by dvd.)

And maybe installing xfce as second choice. I've been using gnome classic, but I was already planning to try something lighter (and more stable), xfce or maté or cinnamon.
(I would have been already on xfce but it wasn't quite enough for my tastes. It has been undergoing active development, so may be ok in mga4. Which I'll be upgrading to in the next few weeks.

And reinstall if necessary.

Thanks again for your feedback. Really appreciated

[doktor5000]

Well, if failsafe mode still work, you could simply try to install kdm or lxdm or xdm and switch to that: http://doc.mageia.org/mcc/3/en/content/drakedm.html
And then see if it starts, that could probably be much less work then reinstalling. Also many of the log messages could be red herrings.
But that's hard to check remotely :/

[andre999]

Thanks for the suggestion. And the pointer to where change the default desktop. I already have XFCE installed alongside Gnome. Maybe that will be enough. If not, I'll try LXDE.
Since my computer is shared (although 99% of the time only me on it), I really want to avoid a new install.

I did manage to squash a few of the messages, but not many.
Some were indeed red herrings, virtually identical to that bug report I mentioned before. (BTW, they are hard to check locally as well )
But in my case, blocked network access also blocked the booting process, which is definitely a bug.

In the process I (re)discovered that I had tried dropbox earlier in january, and had deactivated it not long before my last reboot.
Initially not realising exactly what dropbox was, and how intrusive and bandwidth-hungry it was, even without any shared files.
In looking through the dropbox program folder, I found that it touched all the areas I have error messages and more. Including particularly dbus and network access. It automatically accesses the internet on boot. Which explains the persistant attempts to access the internet, and refusal to continue when it can't. (My internet access is often -- but not usually -- somewhat unstable. particularly on booting.)
Unfortunately something corrupted the process, quite likely my latest mga updates. (Dropbox documentation says they modified a number of standard programs. Probably config files as well.)
So I would suggest avoiding dropbox like the plague. I looked at Ubuntu one as well, and it looks like a much better choice if you need to share files between computers, or with other users.
It looks like dropbox *will* eat your kittens, so to speak.

Thanks again for the feedback. I've noticed that you really do a good job on the forum, when I've consulted it from time to time. I should be helping out here as well, since often I end up giving suggestions to others when looking for answers.
And what better cause than Mageia ?

[doktor5000]

But in my case, blocked network access also blocked the booting process, which is definitely a bug.

Yep, granted. That's because AFAIK all network and/or non-critical shares added by Mageia tools to /etc/fstab miss the _netdev and/or nofail options.
See the following links for more info
https://wiki.mageia.org/en/Mageia_2_Err ... t_in_fstab
https://bugs.mageia.org/show_bug.cgi?id=7673
viewtopic.php?f=25&t=6887
viewtopic.php?f=8&t=6219
viewtopic.php?f=25&t=2549

[andre999]

Your suggestion to change the boot manager worked.
Many thanks for the suggestion and the link given.

I ended up choosing lightdm, which was already installed, after adding lxdm and xdm. (Just to make sure that something would work)
Lightdm has a nice simple interface, and gives me everything I need.

From everything I saw, I'm guessing that dropbox, not having an uninstall option, left some configuration changes that affected gdm/gnome.
(There was an error message with usr/bin/gnome-session and usr/share/gdm in it, not long before the first shutdown message in the boot log, but I had only noticed the dbus and systemd in the same very long line.)
The changes didn't cause a problem until the first reboot I did after an update.
There were already a few things on gnome that didn't work, gdm is just added to the list.

BTW, there was no external network added to my fstab, dropbox seems to do everything from user space after login.

I'm still a little surprised at how well everything works now
Thanks again

[doktor5000]

BTW, there was no external network added to my fstab, dropbox seems to do everything from user space after login.

Yep, it does, no mounts via fstab.