For certain installations of Cloudian, one would want to enable specific Erasure Coding schemes that are not enabled by default in Cloudian Hyperstore.
In my case, I wanted to add the 3+2 scheme (requiring 5 nodes in the cluster).
Even if this is a customizable setting within Hyperstore, it is absolutely advisable to discuss the matter with your local Cloudian technical representative to ensure that you will end up with a supported configuration. Worst case scenario, you end up creating an Erasure Coding scheme that absolutely should not be put in production – and is not supported by the vendor at all. Please do not commit any changes based on this guide without discussing it with Cloudian first.
Now, lets get down to how it’s actually done.
Process
- Log in to your Puppet Master node via SSH. This is typically node #1 but you can verify this from the CMC by viewing the “Cluster > Cluster Configuration” tab.
- Change the directory to
/etc/cloudian-"version"-puppet/modules/cloudians3/templates/ - Edit the
mts.properties.erbfile usingvi - Add the following line at the end of the file
cloudian.protection.policy.ec.schemes=3:2,4:2,6:2,8:2,9:3,12:4 - Launch the Cloudian Installer script from within the current release’s staging directory on the Puppet Master node:
# cd /opt/cloudian-staging/7.2.4/
# ./cloudianInstall.sh - Select option “2” for “Cluster Management”
- Select option “b” “Push Configuration to Cluster” to initate a puppet push on all nodes
- Once the puppet push has successfully completed, return to the menu
- Select option “c” to “Manage Services”
- Select “0” for “All Services”
- Type “restart”
- Once all services have restarted, exit the cloudian installer script.
- Log back in to the CMC as admin and verify you have the option to create an Erasure Coding Protection Policy using your chosen scheme (in my case 3+2).
This job was done on a new Cloudian Hyperstore 7.2.4 cluster, and the process is described with the help of documentation provided from Cloudian.
