Created 10-11-2016 02:30 PM
Is it possible to use cloud break and ambari blueprints to deploy an EMC isilon cluster using openstack?
Can you post an example? For example, can you specify in an ambari blueprint the location of the name node and data node but have ambari skip the install?
Created 10-20-2016 05:56 PM
Hi @cduby,
As requested:
What do you do during manual registration? Do you install additional packages or just configure the data and namenode differently? In the latter case you can put those properties into the blueprint and you can do the installation and configuration in one step. If you are doing e.g. package installation and configuring things outside Hadoop config files the I see two options:
Basic Option (with manual steps):
Advanced Option:
Thanks,
Attila
Created 10-12-2016 03:03 PM
Hi,
Cloudbreak uses Heat templates to set up networking, launch VMs and attach disks to them. The volume management is abstracted under Cinder on OpenStack, therefore it is not relevant for Cloudbreak that what storage backend is used under the hood. Although we have never tested EMC Isilon, but if it can be managed through Cinder then Cloudbreak can handle it.
For your reference I copy here the Heat template fragment used by Cloudbreak, that is responsible to create the necessary cinder volume resources:
type: OS::Cinder::Volume properties: name: hdfs-volume size: ${volume.size}
The complete template is here:
Attila
Created 10-18-2016 02:41 PM
For iSilon, we need to add a node to the cluster using Ambari manual registration after the cluster is provisioned. So first we would need to create a blueprint with no data or name nodes, deploy it on open stack, and then after the cluster is created add the data and name node using manual registration.
It would probably be ok to have the manual registration as a manual step but would be even better if it could be automated.
Created 10-19-2016 07:08 AM
What do you do during manual registration? Do you install additional packages or just configure the data and namenode differently? In the latter case you can put those properties into the blueprint and you can do the installation and configuration in one step. If you are doing e.g. package installation and configuring things outside Hadoop config files the I see two options:
Basic Option (with manual steps):
Advanced Option:
Attila
Created 10-20-2016 04:43 PM
@Attila Kanto This looks like the information I was looking for. Can you post this as an answer and I will accept it?
Created 10-20-2016 05:56 PM
Hi @cduby,
As requested:
What do you do during manual registration? Do you install additional packages or just configure the data and namenode differently? In the latter case you can put those properties into the blueprint and you can do the installation and configuration in one step. If you are doing e.g. package installation and configuring things outside Hadoop config files the I see two options:
Basic Option (with manual steps):
Advanced Option:
Thanks,
Attila
Created 03-10-2017 12:01 AM
@cduby Sorry for posting to this closed issue. I'm interested to know if you were truly able to deploy this way though.
I have never done a Cloudbreak installation to Openstack with Isilon, so I'd love to find out more about what you did.