Please review similar questions on our site, https://community.hortonworks.com/questions/4940/hdp-os-upgradepatching-best-practices.html
We also have documentation on doing OS upgrades, there's a link in the community thread I provided.
Hi @Michael Reyes, we have successfully patched all cluster nodes including firmware upgrade, Java upgrade etc. while keepint the cluster up and runnig. It assumes HA of major components like HDFS, RM and HBase, and rack awareness of workers. In such a cese you can upgrade masters one by one, upgrade workers rack by rack and Kafka brokers node by node. Of course, it's good to test first in a Dev or Test envirnonment.