How logs and data will be managed by mapreduce history server when two instance are running in HA mode?

[#1]

I have a 3 node setup and want to achieve high availability for map reduce history server. To do so i have installed history server on two nodes and started both of them and managing them with a virtual ip. I have tested it for a oozie workflow and its working fine. I want to know that how it going to be manage logs and data?

[#2]

This little dated articles may help https://hortonworks.com/blog/simplifying-user-logs-management-and-access-in-yarn/ and https://hortonworks.com/blog/elephants-can-remember-mapreduce-job-history-in-hdp-2-0/ . Also take a look at the timeline server doc. May also be worth taking a look at Tez timeline server if you are using Tez for your jobs. http://docs.hortonworks.com/HDPDocuments/HDP2/HDP-2.4.3/bk_yarn_resource_mgt/content/ch_timeline_ser...

https://tez.apache.org/tez-ui.html