Hi,

we are starting to think about the upgrade of our existing CDH 4.2.0  Cluster 
to CDH5.0.1

We have a master Hadoop cluster with hbase and a slave Hadoop cluster with hbase
Hbase replication is enabled from master to slave cluster.

What is the best was to upgrade this environment with a minimum downtime of the 
master cluster?

My suggestion since now is this way.  

1.) shut down hbase in the master cluster

2.) shut down hbase on the slave cluster

3.) shut down hadoop on the master cluster

4.) upgrade hadoop on the master cluster

5.) upgrade zookeeper server for master cluster

6.) upgrade zookeeper server of the slave cluster

7.) start hadoop on the master cluster

8.) upgrade hbase on the master cluster

9.) start hbase on the master cluster

10.) shut down hadoop on the slave cluster

11.) upgrade hadoop on the slave cluster

12.) start hadoop on the slave cluster

13.) upgrade hbase on the slave cluster

14.) start hbase on the slave cluster

Did i miss a step? 
If so which step is missing and at what position?

After these steps will replication start automatically?
Or do we have to start the replication again and have to run
copytable jobs for the timerange of the upgrade?

Regards Hansi

Reply via email to