[CentOS] Still won't boot - Re: remote yum update aborted with session closing - won't boot

Wed Jul 2 14:45:34 UTC 2014
Robert Moskowitz <rgm at htt-consult.com>

On 07/02/2014 10:33 AM, Reindl Harald wrote:
>
> Am 02.07.2014 16:31, schrieb Reindl Harald:
>>
>> Am 02.07.2014 16:27, schrieb Reindl Harald:
>>> Am 02.07.2014 16:20, schrieb Robert Moskowitz:
>>>> On 07/02/2014 09:35 AM, Reindl Harald wrote:
>>>>> Am 02.07.2014 15:19, schrieb Robert Moskowitz:
>>>>>> On 07/02/2014 08:41 AM, Reindl Harald wrote:
>>>>>>> Am 02.07.2014 14:32, schrieb Robert Moskowitz:
>>>>>>>> On 07/01/2014 06:25 PM, Frank Cox wrote:
>>>>>>>>> On Tue, 01 Jul 2014 18:19:32 -0400
>>>>>>>>> Robert Moskowitz wrote:
>>>>>>>>>
>>>>>>>>>> How do I trouble shoot this?  I am assuming that I only got a partially
>>>>>>>>>> completed update.
>>>>>>>>> Try yum-complete-transaction and see what happens.
>>>>>>>> Well it looks like all the updates 'took'  /etc/redhat-release now
>>>>>>>> reports ver 6.5.
>>>>>>>>
>>>>>>>> But still getting a kernel panic and boot failing.  I can fall back to
>>>>>>>> the prior kernel and it will boot.  So I tried to copy the lines off the
>>>>>>>> monitor, as nothing is getting logged:
>>>>>>>>
>>>>>>>> IOMMU: failed to mpa dmn 0
>>>>>>>> kernel panic - not syncing: VFS: Unable to mount root fs on
>>>>>>>> unknown_block (0,0)
>>>>>>>> Pid:1, comm swapper Tainted: G  -------------} 2.6.32-431.20.3.el6.i686 #1
>>>>>>>>
>>>>>>>> then a a number of dump lines.
>>>>>>>>
>>>>>>>> The kernel that is working (only other one listed in menu) is:
>>>>>>>> 2.6.32-279.22.1.el6.i686
>>>>>>> why don' t you just remove any package from the new kernel
>>>>>>> which don't work anyways and after that "yum upgrade" again?
>>>>>> But how do I determine what package from the new kernel
>>>>> i said: and after that "yum upgrade" again man if you remove that
>>>>> broken packages a "yum upgrade" pulls them simply as if they where
>>>>> never be installed and the rpmscripts will generate a new initrd
>>>> OK.  So...
>>>>
>>>> # yum erase kernel*
>>>>
>>>> Transaction Summary
>>>> ===========================================================================================================
>>>> Remove       17 Package(s)
>>>>
>>>> So I am making notes about what is being removed so I can insure that it
>>>> gets reinstalled with the upgrade
>>> no you remove only the broken kernel and not a blind "+"
>>>
>>> rpm -qa | grep kernel
>>> uname -r
>>>
>>> yum supports to be specific
>>> example: "yum remove kernel-2.6.32-431.20.3.el6.x86_64"
>>>
>>> the kernel package itself is the one shipping drivers
>>> and trigger rebuild of the initrd
>>>
>>> look that the dependencies is pulling - most likely "kernel-firmware"
>>>
>>>> Well the upgrade is NOT reinstalling all that was erased
>>>> So looks like I will have to do all the others step by step
>>> because you asking, then doing and then asking how to proceeed instead
>>> consider "hmm remove LSB and GLIBC parts is not a godd idea"
>> rpm -e --nodpes \
>>   kernel-2.6.32-431.20.3.el6.i686 \
>>   kernel-headers-2.6.32-431.20.3.el6.i686 \
>>   kernel-firmware-2.6.32-431.20.3.el6.i686
>>
>> yum upgrade
> boah damned - "rpm -e --nodeps" instead "--nodpes"
>
> that removes only the named packages *but* be sure the version
> is really the non-bootable and in that case the broken
> dependencis don't matter, "yum upgrade" will pull the new
> kernel again with all it's deps
>
> that way you avoid to remove the dependency chain, however
> you already removed it instead wait for a response after
> asking questions
>
As typical, you are right, Harald.  I have a conference call in 15min 
where one of the discussions will be putting up my Teredo testbed, and 
this system is what is handy for the Teredo (Miredo) server.  So I 
rushed more than I should have.  But I did piece it all together.

Now I can pull the rpmforge Miredo code and get that up.  Too bad that 
it is not in EPEL 6; the Miredo maintainer has not commented on why it 
is not there.