[one-users] VM sticks in 'pend'

[Dan Yocum]

Tino,

On 06/18/2010 11:36 AM, Tino Vazquez wrote:
> Hi Dan,
>
> comments interspersed,
>
> On Fri, Jun 18, 2010 at 1:20 PM, Dan Yocum<yocum at fnal.gov>  wrote:
>> Tino,
>>
>> Thanks for the reply.  I also IM'd with jfontan yesterday morning and got
>> some things worked out.  Yes, you're correct, the scheduler was not running.
>>   I was starting 'oned -f&' manually, and not using 'one start'.
>>
>> After a little back and forth I was able to use the Opennebula Express
>> install as a basis to run a VM under KVM on our Scientific Linux 5.4 (which
>> is RHEL, really) systems.  Some hacks had to be performed, but that's the
>> nature of the game.  I also ran into some discrepancies in the documentation
>> - where should I send the feedback to?
>
> Glad to hear that. You are welcome to send any feedback you feel like
> to the users mailing list or directly to me if you want.
>>

I will collect my notes and send them to you and the list.

>> Anyway, now I've got a VM running on a worker node, but it apparently can't
>> see the network and I can't ssh or 'virsh console' into it.  Any
>> suggestions?
>>
>> Here's some output:
>>
>> $ onevm show 10
>> VIRTUAL MACHINE 10 INFORMATION
>> ID             : 10
>> NAME           : fgitb318vm0
>> STATE          : ACTIVE
>> LCM_STATE      : RUNNING
>> START TIME     : 06/17 14:34:36
>> END TIME       : -
>> DEPLOY ID:     : one-10
>>
>> VIRTUAL MACHINE TEMPLATE
>> CONTEXT=[
>>   FILES=/srv/cloud/one/one-templates/init.sh /srv/cloud/one/.ssh/id_dsa.pub,
>>   HOSTNAME=fgitb318vm0,
>>   IP_PUBLIC=131.225.154.24,
>>   ROOT_PUBKEY=id_dsa.pub,
>>   TARGET=hdc,
>>   USERNAME=opennebula,
>>   USER_PUBKEY=id_dsa.pub ]
>> CPU=0.1
>> DISK=[
>>   READONLY=no,
>>   SOURCE=/srv/cloud/one/one-templates/ttylinux.img,
>>   TARGET=hda ]
>> FEATURES=[
>>   ACPI=no ]
>> MEMORY=64
>> NAME=fgitb318vm0
>> NIC=[
>>   BRIDGE=br1,
>>   IP=131.225.154.24,
>>   MAC=00:03:83:e1:9a:18,
>>   NETWORK=Small network,
>>   VNID=0 ]
>> VMID=10
>>
>> I've had a VM running under KVM at this IP in the past with no problems.
>
> Ok, I assume you are using the ttylinux example we provide, which is
> already contextualized (the vmcontext.sh init script that takes the
> NIC's MAC address, derives the IP and sets it). In order to see what
> is going on inside the VM, the best thing is to use VNC. Add this to
> the VM template:
>
> GRAPHICS = [
>    type    = "vnc",
>    listen  = "0.0.0.0",
>    port    = "5900"]
>
>
> and connect to the port 5900of the physical host where the VM is
> running using a VNC client.


I migrated my test instance to another set of machines and am now back 
to this point.  I was able to attach the vncviewer to the image and lo 
and behold, it said:

This filesystem has gone 302 days without being checked.  Check forced.

****** An error occured! Dropping to a root shell.
****** Type the root password at the prompt.

Of course, typing the root password and running fsck on the filesystem 
revealed no errors, so I rebooted and it came up.

That said, whoever created the ttylinux tarball should really go in 
there and turn off filesystem check on the image so future people 
following the instructions don't run into this problem.  Running 
'tune2fs -i 0 /dev/hda1' will fix the problem.

Onward!

Thanks,
Dan

-- 
Dan Yocum
Fermilab  630.840.6509
yocum at fnal.gov, http://fermigrid.fnal.gov
"I fly because it releases my mind from the tyranny of petty things."