Updating long-running clusters
Long-running clusters require maintenance updates such as updates to the base image's operating system and any third party packages that have been installed. Cloudbreak supports performing these maintenance updates.
Ambari and HDP/HDF stack updates are not supported.
- Preparing a new custom image (if using custom images).
- For each running cluster, manually updating existing nodes to use the same versions as the new image.
- For each running cluster, setting that new image as default for all newly added nodes. The new default image will be used for the new nodes added to the cluster.
This is illustrated in the following diagram:
The following limitations apply:
You can only upgrade to a pre-warmed image if the Ambari/HDP/HDF versions are the same on the new image and on the old image.
For base image only, the Saltboot and Salt versions must match on the new image and on the old image.
In all cases, Linux OS must be the same. For example, you cannot upgrade from Redhat to Centos. Furthermore, the major X version must also be the same (i.e. you cannot update from version 6 to 7, but you can update from 6.1 to 6.2).