The errors I'm seeing in the logs relate to virtualization required baked into the AMI for the large memory instances. This is an error on our side related to new instance types.
What I'd recommend to get off the ground immediately (with reasonable memory on a node) would be to use the c3.8xlarge instance type (compute optimized 8x, 60GB memory), which will launch successfully.
On Fri, Oct 10, 2014 at 1:35 PM, Chris Dagdigian dag@sonsorol.org wrote:
Appreciate the fast reply! I'm at a training session where I hope to spin up 24 cloudman managed galaxy instances so each student can have a 1:1 ownership of their own cluster
I used this URL to fire up my test cluster:
https://usegalaxy.org/cloudlaunch
... but the test cluster is running out of RAM building bowtie2 index files so I was trying to start a new cluster based on the currently running system but started on an instance type with more memory.
And today when I go back to the https://usegalaxy.org/cloudlaunch I'm getting the errors I posted
If there is anything in the logs I'd appreciate it. Not sure if I'm doing anything wrong on my end.
Currently launching the pure base AMI without the cloudlaunch interface to see if I get further ...
Regards, Chris
Dannon Baker wrote:
Just realized you said cloudlaunch in the subject there, so I'm assuming that's what you used. For both working and non-working clusters? And this is cloudlaunch on usegalaxy.org http://usegalaxy.org? I can check the logs for more detailed launch failures, if so.