Bug 680332

Summary: blogd crashes on boot and "kills" the runlevel initialisation
Product: [openSUSE] openSUSE 11.4 Reporter: kolA flash <kolAflash>
Component: OtherAssignee: E-mail List <bnc-team-screening>
Status: RESOLVED DUPLICATE QA Contact: E-mail List <qa-bugs>
Severity: Normal    
Priority: P5 - None CC: arne.reiter, forgotten_-QHEa95TGX, jes, werner
Version: Final   
Target Milestone: ---   
Hardware: x86-64   
OS: Other   
Whiteboard:
Found By: --- Services Priority:
Business Priority: Blocker: ---
Marketing QA Status: --- IT Deployment: ---

Description kolA flash 2011-03-16 23:13:03 UTC 
User-Agent:       Mozilla/5.0 (X11; Linux x86_64; rv:2.0b12) Gecko/20110222 Firefox/4.0b12

After updating to 11.4 (x86_64 - from 11.3) on a AMD Phenom II pc I got this error at the end of the boot-log:

blogd:Can not read from fd 0: Input/output error

Seems like blogd crashes. And by doing this, it stops the system from entering the correct runlevel. "fd 0" seems to be the filehandle 0. But I'm not shure that this is the meaning of "fd 0".
There are no messages about the services that should be startet in that runlevel.
And for example the X-Server doesn't comes up in runlevel 5 (so no XDM, KDM or XOrg shows up).





If I log in after boot-up, and type "runlevel" it tells "N 5".
For debugging, I put some code into /etc/init.d/boot.local, which checked the filehandles of blogd. The result:

ls -lA /proc/`pidof blogd`/fd/
total 0
lrwx------ 1 root root 64 Mar 16 23:34 0 -> /dev/ptmx
l-wx------ 1 root root 64 Mar 16 23:34 1 -> /dev/tty1
l-wx------ 1 root root 64 Mar 16 23:34 2 -> /dev/tty1
lrwx------ 1 root root 64 Mar 16 23:34 3 -> /dev/blog
l-wx------ 1 root root 64 Mar 16 23:34 4 -> /var/log/boot.msg


The permissions of /dev/ptmx are:
crw-rw-rw- 1 root tty 5, 2 17. Mär 00:07 /dev/ptmx





Renaming /sbin/blogd to /sbin/blogd.disable seems to be a workaround. After doing this, my system initializes the runlevel as it should.




I found just 2 similar problems using Google:
http://forums.opensuse.org/english/get-technical-help-here/hardware/455513-opensuse-11-4-nvidia-install-post2306191.html
http://linux.derkeiler.com/Mailing-Lists/SuSE/2011-02/msg00490.html

Reproducible: Couldn't Reproduce

Steps to Reproduce:
Difficult to reproduce. I did the pretty same update procedure on my notebook - but there everything's working fine.
Actual Results:  
Any service of the choosen runlevel comes up. No xdm, kdm, xorg.

Expected Results:  
Runlevel should initialise correctly and blogd should be able to do whatever it should do.

Workaround: Rename or delete /sbin/blogd (for example into /sbin/blogd.disable)

2 similar problems:
http://forums.opensuse.org/english/get-technical-help-here/hardware/455513-opensuse-11-4-nvidia-install-post2306191.html
http://linux.derkeiler.com/Mailing-Lists/SuSE/2011-02/msg00490.html
Comment 1 John Sheehy 2011-03-17 00:49:18 UTC 
I've been bit by this. It prevents the box from booting. I think it should be a higher severity (critical).

It does not occur consistently. It's like playing the wheel of misfortune. 

Here's the last of the info from my boot.log before the machine hung due to blogd.

System Boot Control: Running /etc/init.d/boot.local
done
<notice -- Mar 16 20:11:34.660993000> 
killproc: kill(429,3)

Boot logging started on /dev/tty1(/dev/console) at Wed Mar 16 20:11:34 2011
Master Resource Control: previous runlevel: N, switching to runlevel: 5
Master Resource Control: Running /etc/init.d/before.local
done
<notice -- Mar 16 20:11:34.872409000> service earlysyslog start<notice -- Mar 16
 20:11:34.886670000> service dbus start<notice -- Mar 16 20:11:34.967933000> ser
vice acpid start

--------

/etc/init.d/boot.local is empty.
Comment 2 Forgotten User -QHEa95TGX 2011-03-19 16:29:46 UTC 
see also https://bugzilla.novell.com/show_bug.cgi?id=642289
Comment 3 kolA flash 2011-03-19 21:33:57 UTC 
Hmm... looks like #642289
BUT #642289 is marked as duplicate of #642334 and #642334 DOES NOT LOOK LIKE THIS/MY PROBLEM!!!

BEFORE YOU ANYONE SETS THIS AS A DUPLICATE OF #642289 MARK #642289 AS >>>NOT RESOLVED<<< !!!

Thanks :-)
Just want to make sure, my bug doesn't gets lost :-)
Comment 4 kolA flash 2011-03-20 00:21:46 UTC 
My notebook just started to show the same bug after installing some patches (via YaST). The notebook is also updated from openSUSE 11.3 (like my pc, where I experienced the bug the at first time). But the notebook has a 32 bit openSUSE.

The patch which looks the most suspicious is for tha package "at".
The other patches are for packages like Java or MPlayer. So they shouldn't be responsible for such a system-near bug.



... If I keep thinking a little longer: "at" shouldn't be responsible for breaking "blogd". It just doesn't makes sense. "at" is for job scheduling. It shouldn't break a tool logging the system boot.

I didn't used my notebook since upgrading it from openSUSE 11.3 to 11.4. So maybe the bug already existed on my notebook before (since upgrading to 11.4), but just doesn't shows up every time.
Comment 5 Dr. Werner Fink 2011-03-21 14:47:58 UTC 
It seems also a problem taht somehow the fd 0 of blogd is stolen

*** This bug has been marked as a duplicate of bug 642289 ***