[Mageia-dev] Fwd: Re: M3 won't complete boot after update

Colin Guthrie mageia at colin.guthr.ie
Thu Sep 27 11:15:51 CEST 2012


'Twas brillig, and Anne Wilson at 27/09/12 09:46 did gyre and gimble:
> On 26/09/12 16:12, Marja van Waes wrote:
>> Can you try doing "systemctl show prefdm.service | grep 
>> ActiveEnter.*Mono" after a fresh boot.
> 
>> I don't need to know the results, just whether it's non-zero or 
>> still at 0.
> 
>> Assuming it's still at zero, then we've not even tried to start the
>> graphical server.
> 
>> I suspect this is because we're considering your remote mounts as 
>> critical to the user sessions (i.e. think of /home on NFS). 
>> Normally, putting nofail in the fstab is enough to break this 
>> linkage.
> 
>> Can you comment out all remote mounts entirely and see if you can 
>> boot.
> 
>> If it still fails, then try starting prefdm.service manually 
>> (systemctl start prefdm.service) and see what happens.
> 
> At this point I'm probably stuck.  I've been doing these tests by
> forcing a level 1 boot, but now even that may not be possible.  These
> are the latest boot messages:

Does it work when forcing a runlevel 3 boot instead. This is closer to a
proper boot and likely a better testing environment. I presume the
getty's are showing for you in level 3?

> broadcom-wl (....)Already exists on this kernel [OK]
> Checking for new hardwareWelcome to rescue mode! Type "systemctl
> default" or ^D to enter default mode ##Yes, that's how it looks, lines
> joined.

Interesting that they are joined up...

> Type "journalctl" to show system logs. Type "systemctl reboot" to
> reboot. [OK]
> ##didn't do any of these as I was not watching the screen
> 
> Failed to issue method call: Transaction is destructive.

I've seen this before, but it's basically indicative of some kind of
transaction ordering deadlock. I'll investigate to see if I can
reproduce that here.

I have seen it when you have SINGLE=/sbin/sulogin rather than
SINGLE=/sbin/sushell in /etc/sysconfig/init.

> Last resort -
> 
> I edited out the "splash quiet" and called for level 1 again.  This
> time it completes, very quickly.
> 
> 'systemctl show prefdm.service | grep ActiveEnter.*Mono" ' simply
> returns to the prompt.

This shouldn't happen. These should be printed even if they have zero
values I believe, although with a level 1 or level 3 boot this might not
be the case (I've not tested). It would certainly *always* be 0 if they
are shown tho',

The results of this test are only really valuable if you do a normal
boot to runlevel 5 as this is when the service should be started.

What I'm trying to work out here is if the job relating to starting the
prefdm.service had to be ejected due to some kind of conflict.

If there is a conflict then the values will be 0 and this will explain
why your graphical desktop does not start.

> All the borg2 mounts are commented out for now.
> 
>> If it still fails, then try starting prefdm.service manually 
>> (systemctl start prefdm.service) and see what happens.
> 
> Nothing at all happens.  It returns to the prompt.

This is curious. Does this also occur at runelevel 3? What does
systemctl status prefdm.service tell you after it returns you to the prompt?

Is there anything useful in the kdm log file about why kdm failed to
launch X11?

> We have all spent a lot of time on this.  There are files on the
> netbook that I must recover, so now it's decision time.  It would
> probably be easier and quicker to grab the latest image (in which case
> can you remind me where to go for that) but I am willing to continue
> if there is any possibility that what we discover could save grief for
> others.
> 
> What do you think?

Plenty more to debug if you want to continue.

Col




-- 

Colin Guthrie
colin(at)mageia.org
http://colin.guthr.ie/

Day Job:
  Tribalogic Limited http://www.tribalogic.net/
Open Source:
  Mageia Contributor http://www.mageia.org/
  PulseAudio Hacker http://www.pulseaudio.org/
  Trac Hacker http://trac.edgewall.org/


More information about the Mageia-dev mailing list