FIX - Failure prepping block device Error - Openstack

Joshua Holmes

4/5/15

So, big news, Ethode has been working on ramping up an OpenStack cluster (juno) in our Medina office location, using Ubuntu 14.04 as our primary Operating system. Our current plans will end up with close to 1000 CPU cores, 50TB of Storage and 20+ Compute nodes across a 10GB storage network. We ran into an issue setting up a new Storage node with 2 single 1TB SSD drives and 8 CPU cores.

There were two errors for the same process.

1) Failure prepping block devices

and

2) TypeError: create_volume() got an unexpected keyword argument 'source_replicaid'

Both of these errors are failure ambiguous specially if you're new to OpenStack. Fortunately the second error message immediately gives a hint into the problem. Anytime I see issues where API calls seem to be missing data, it's always plausible that the versions of the software you're running do not match.

In this case, the problem was simple to solve. The controller node is running the Cinder API software and your storage node is running Cinder Volumes software which talks to the controller.

If your versions are not near the same release time frames it chokes. When you setup OpenStack in my case Juno - you MUST make sure and add the Cloud repository to EACH server and run a full OS update before proceeding to install the required software otherwise you wind up with what ever it latest approved for your distro. Fortunately, THE FIX IS EASY. Run this process ON ALL SERVERS IN YOUR CLUSTER THAT YOU HAVEN'T ALREADY DONE SO ON.

apt-get install ubuntu-cloud-keyring <hit enter>

echo "deb http://ubuntu-cloud.archive.canonical.com/ubuntu" \
  "trusty-updates/juno main" > /etc/apt/sources.list.d/cloudarchive-juno.list <hit enter>

apt-get update && apt-get dist-upgrade <hit enter>

Once you do this Ubuntu will download the correct packages from the Canonical Cloud packages repository and you will have the correct version of Cinder volumes (storage node) and cinder API (controller)