According to the release note of StarCluster 0.95, it says now it supports g2.2xlarge node.
So I've tried StartCluster 0.95 to launch following AMI on (G2 GPU) node at Oregon region,
 ami-80bedfb0 us-west-2 starcluster-base-ubuntu-13.04-x86_64-hvm (HVM-EBS)
 ami-486afe78 us-west-2 starcluster-base-ubuntu-12.04-x86_64-hvm (HVM-EBS)
but got an error like
Instances of type 'g2.2xlarge' may not be launched from AMI 'ami-486afe78'
(Also same in other regions)
Is StarCluster's AMI not yet ready for G2 GPU node? or is it the problem of EC2?
@syoyo My apologies for the delay. I'm currently in contact with AWS to fix this for our HVM AMIs. I'll close this when it's working. Stay tuned.
Cool. I'll wait. Thank you.
Is there any update regarding that issue?
Also trying to do the same. Any update?
Also, what is the magic that makes some AMIs work and others not for this instance type? It doesn't seem to be just the HVM. For example, I tried the following one and also got the same error:
starcluster-base-ubuntu-12.04-x86_64-hvm - ami-52a0c53b
StarCluster Base Ubuntu AMI 12.04 x86_64 HVM/CC/GPU (US-EAST-1)
I'm also trying to get a g2.2xlarge to boot in us west 2c using starcluster-base-ubuntu-12.04-x86_64-hvm ami-1db8282d. This produces: Launch Failed
Instances of type 'g2.2xlarge' may not be launched from AMI 'ami-1db8282d'. I was doing this directly in the AWS console. I was doing this as a sanity check (this works for other starcluster AMIs I have when running on c3.large). Then if that worked I was going to use starcluster to start it up and configure it.
Is it possible that the following applies to the above public AMI?
See the section: AMIs for GPU Instances
Client.UnsupportedOperation: Instances of type 'g2.2xlarge' may not be launched from AMI .
From the AWS forums https://forums.aws.amazon.com/thread.jspa?messageID=539055򃦯
It seems that the AMIs have to be GPU Backed in order to launch g2.2xlarge instances.
Things seem to launch fine with current public starcluster AMIs.