willmore / d2c Goto Github PK

Write method that takes a FS file and:

1. Adds EC2-suitable virtual kernel
2. Modifies /etc/fstab to conform with EC2 instance format

When creating instances, EC2 can reject requests for various reasons. For example: not enough spare instances available.
Capture this exception and properly stop the deployment, de-allocating any active instances.

Example trace:
File "44_py_run.py", line 87, in main
deployment.run()
File "/usr/local/lib/python2.6/dist-packages/d2c-1.0-py2.6.egg/d2c/model/Deployment.py", line 170, in run
transition()
File "/usr/local/lib/python2.6/dist-packages/d2c-1.0-py2.6.egg/d2c/model/Deployment.py", line 185, in __launchInstances
role.launch()
File "/usr/local/lib/python2.6/dist-packages/d2c-1.0-py2.6.egg/d2c/model/Role.py", line 76, in launch
instance_type=self.instanceType.name)
File "/usr/local/lib/python2.6/dist-packages/boto-2.0b4-py2.6.egg/boto/ec2/connection.py", line 575, in run_instances
return self.get_object('RunInstances', params, Reservation, verb='POST')
File "/usr/local/lib/python2.6/dist-packages/boto-2.0b4-py2.6.egg/boto/connection.py", line 611, in get_object
raise self.ResponseError(response.status, response.reason, body)
boto.exception.EC2ResponseError: EC2ResponseError: 400 Bad Request

InstanceLimitExceededYour quota allows for 5 more running instance(s). You requested at least 7d502e550-ea35-4a70-bffe-f720343e3196

http://virt-tools.org/
http://libvirt.org/python.html

Create a python script (commit perhaps to test directory) which demonstrates creating and launching a Ubuntu VirtualBox instance. Once the instance launches, the script should SSH into the instance and do something trivial (run a simple bash script maybe). The python script should then shutdown the instance and de-register it from VirtualBox.

When creating an AMI, connection to EC2 with boto library when network is down takes too long to timeout. Try reduce this to 3sec and produce meaningful exception.

Fix field sizes, margins, font sizes, etc.

In the GUI, the user should not be able to define a role such that the image and instance architecture do not match.

SciCloud has some instances with disk too small for the root image. Prevent attempted launch of this.

According to http://wiki.xensource.com/xenwiki/XenOverview
kernels since 2.6.24 already support Xen PV, so there should be no need to update images if the kernel.

In the image EC2izing process:
If an image already has a PV capable kernel, we should not install a new kernel and modules.
If the cloud does not support PV grub, we should create a new AKI and ARI on-the-fly and attach to the image.

If exception occurs while running, allow user to either:

1. Terminate
2. Restart from current state (after they have maybe fixed something manually)

VMs must have sshd installed to be usefull. Detect this during the AMI creation phase. If not installed - report exception and abort.