Chameleon Cloud provides OpenStack Cloud with KVM or Baremetal for developing with the Big Data Stack. https://www.chameleoncloud.org/ Chameleon Cloud provides an environment for experimenting with cloud environments. This class does not the support use of Chameleon cloud for any assignment or project work but computing resources may be available upon request with limited access and allocation.
There are some differences between FutureSystems OpenStack and Chameleon OpenStack.
- different login usernames (
cc
for all instead ofubuntu
,centos
, etc) for the images- limited resource availability
- resource usage is charged to a finite allocation (thus you need to terminate your instances if you do not use them).
- Create an account on [[https://www.chameleoncloud.org/][Chameleon Cloud]]
- Send your Chameleon username to <course email> Note: the subject MUST be #+BEGIN_EXAMPLE Chameleon Cloud Access #+END_EXAMPLE
- We will then add you to the project for this course. IMPORTANT: you will
- not be able to use Chameleon until you are added. We will reply to your request with a confirmation email.
ssh into
india.futuresystems.org
- Go to the
Chameleon Cloud OpenStack Dashboard and download the openrc file (check under the
API Access
tab)Upload the openrc file to the
india
node:: $ scp CH-817724-openrc.sh $PORTALID@india.futuresystems.org:~
- Upload your india ssh key to your `profile on github.com
albert@i136 ~ $ cat ~/.ssh/id_rsa.pub
Source the openrc file (only the chameleon openrc file):: albert@i136 ~ $ source ~/CH-817724-openrc.sh
Load the OpenStack module (same as with kilo on india):: albert@i136 ~ $ module load openstack
At this point you the nova commands will control Chameleon.
You can now follow the Big Data Stack Readme for starting and deploying your BDS
apt
related errors
You may occasionally get an error when one of the tasks calls to apt, either to update the cache or install packages. This will likely manifest as aFailed to fetch
with anError 403 Forbidden
error. The root cause for this is not yet known, but it seems related to a network saturation issue. Nonetheless, the workaround is simple: rerun the playbook that failed. This may need to be repeated a few times, but this has been sufficient to resolve the issue when I encounter them.