[StarCluster] starcluster AMI that can boot g2.2xlarge on AWS?

[Silverstein]

Hi,

I'm looking for a starcluster AMI that preferably has cuda built-in and 
that can boot on a g2.xlarge on AWS. Being able to boot on g2.xlarge is 
the primary requirement.

I found the following AMI: starcluster-cuda-small-ami (ami-580b6e68).  
However, when I started it via starcluster it hung at the ssh 
configuration stage:

>>>  Waiting for all nodes to be in a 'running' state...

1/1 |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| 
100%

>>>  Waiting for SSH to come up on all nodes...

0/1 | |   0%


After figuring out how to log in to the node that was now running (since 
the normal starcluster accounts were not accessible) I discovered it 
cranking away on the GPUrunning jhProtominer. Apparently this is some 
sort of bitcoin miner trojan.

This was surprising as I wasn't expecting anything to be running let 
alone something like this.  It was also disappointing because it means 
that I can't use this AMI.  And it does make me now suspicious of AMIs 
including starcluster AMIs (maybe I should have been more so all along).

So my questions are:
1. where can I find a starcluster AMI that can boot on a g2.xlarge (I 
tried the two HVM AMIs' listed by 'starclsuter listpublic' -- 
ami-486afe78 and  ami-80bedfb0 -- but they both failed to boot with:  
ERROR - UnsupportedOperation: Instances of type 'g2.2xlarge' may not be 
launched from AMI.)
2. how are the AMIs available on the stcluster site vetted?  How can I 
ensure that these are safe?
3. If there is no way to verify that any given starcluster AMI is safe, 
how do I build a g2.xlarge bootable starcluster AMI as there seem to be 
some specific requirements for this sort of instance.

Thanks,

Herc

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.mit.edu/pipermail/starcluster/attachments/20140522/af37659f/attachment.htm