Ubuntu Enterprise Cloud. VM is pending and then terminating. Why?

I'm trying to setup a private cloud. I followed the tutorial from:

https://help.ubuntu.com/community/UEC/CDInstall

But I'm not able to run a new image. Every time I try to start an image, the VM goes into the pending state and after a while it moves into shutting down mode a开发者_开发百科nd finally it terminates.

This behaviour is reported in the official doc for Eucalyptus, where it suggests to have a look to httpd-nc_error_log (the one for the node controller):

http://open.eucalyptus.com/wiki/EucalyptusTroubleshooting_v1.5.2

I would really like to have a look to this file, but I can't, since nodes are managed by the controller and apparently I don't have ssh access to them (I tried to connect from the controller, but it prompts for a passphrase).

In all other log files (in the controller), I didn't notice anything special.

When I start a VM instance with the following:

euca-run-instances -k mykey emi-**********

The httpd-cc_error_log says:

> RTNETLINK answers: File exists
> Internet Systems Consortium DHCP
> Server V3.1.2 [...] WARNING: Host
> declarations are global.  They are not
> limited to the scope you declared them
> in. Wrote 0 deleted host decls to
> leases file. Wrote 0 new dynamic host
> decls to leases file. Wrote 0 leases
> to leases file. Listening on
> LPF/eth0/**:**:**:**:**:**/euca
> Sending on  
> LPF/eth0/**:**:**:**:**:**/euca
> Sending on  
> Socket/fallback/fallback-net Using
> intrapositioned negation (`--option !
> this`) is deprecated in favor of
> extrapositioned (`! --option this`).
> Using intrapositioned negation
> (`--option ! this`) is deprecated in
> favor of extrapositioned (`! --option
> this`).

Any ideas? Any place where I should look?

I had this problem, the KVM hypervisor requires hardware support for virtualization . Make sure your prossesor supports virtualization and this is activated in the BIOS.

Also, to be able to ssh into your nodes, you have to set up a password on them:

sudo passwd eucalyptus

And after you are done and do not need to log into the node controller, remove the password:

sudo passwd -d eucalyptus

There are a number of reasons for your problem with instances:

1. You have insufficient resources to launch instances.

2. You are choosing the wrong size for your instance. For example, if the image is 2GB in size, you cannot run an m1.small instance, you should go for c1.medium instead. Even though m1.small is 2GB, in reality, it needs extra space.

3. As mentioned above, your processor does not support hardware virtualization (VT) or VT-support is not turned on in BIOS. The node controller has to be created on a machine that has VT support enabled. If you're setting this up on virtual machines, you might also need to direct your hypervisor to make use of VT support.

Check the /var/log/eucalyptus/nc.log on the NC. That's a good place to look for errors like that one.

It's possible that you're experiencing the reported error for Eucalyptus 1.6.2 (or close to that release) where the scsi drivers are not working. You can work around problem by changing all references to sda for hda int the /usr/share/eucalyptus/gen_kvm_libvirt_xml file.