missing image_id parameter for created instances through the ec2 backend

[GoogleCodeExporter]

There is a problem with boto library in the ec2 backend. For some reason it 
doesn't set image_id parameter for created instances, which makes it impossible 
to submit a job to a VM. In the submit_job function the condition:

if (vm.image_id != image_id or vm.instance_type != instance_type):

is never false (because vm.image_id is empty), then the VM is simply skipped 
and gc3pie spawns a new VM, then the story repeats.

(By the way, in the openstack backend, in the corresponding place, there is a 
check only for the image_id and not for the instance_type.)

I am using boto 2.20.1 because the newest version 2.36 (the same with 2.35) 
fails on the boto.connect_ec2() with AuthFailure while connecting to our 
OpenNebula. Absolutely no idea why.
I am using the development branch, revision 4123.

Original issue reported on code.google.com by Karanda...@gmail.com on 2 Feb 2015 at 2:38

[GoogleCodeExporter]

Antonio, can you please have a look at this?

Original comment by riccardo.murri@gmail.com on 10 Feb 2015 at 9:11

[GoogleCodeExporter]

Hi Karandash8,

I am unable to replicate the issue with the given information

In order to debug the issue, we would need the full debug log of the 
application.
If you are running a GSessionBasedScript, run it with -vvv and send us the 
output.
If you are instead running a standalone application, set the logging level to 
DEBUG with something like:

  import logging
  gc3libs.configure_logger(logging.DEBUG, "myapp")

Also, it would be very helpful if you could send us details on the operating 
system you are using and the list of all the PyPI packages installed (output of 
`pip list` command from within the gc3pie virtual environment)

Original comment by arcimbo...@gmail.com on 23 Feb 2015 at 5:12