Kylin Cluster Mode. I am using Kylin on EMR.
Is a restart of kylin required on all job and query servers when new node is
added. Then there will be downtime. How do we achieve High Availibility.

from the guide:
"
Configure the Kylin node list kylin.server.cluster-servers, including all
nodes (the current node is also included). When the event changes, the node
receiving the change needs to notify all other nodes (the current node is
also included).
"

So if I have 2 job servers and 2 query servers:
then
kylin.server.cluster-servers=jobnode1:7070,jobnode2:7070,querynode1:7070,querynode2:7070

If let's say query node 2 on EMR fails. And I add a new node with new EMR
which will have a new IP. Then for the new node:

kylin.server.cluster-servers=jobnode1:7070,jobnode2:7070,querynode1:7070,querynode3:7070

And then looks like I have to change the string on all the nodes to:
kylin.server.cluster-servers=jobnode1:7070,jobnode2:7070,querynode1:7070,querynode3:7070

And then restart kylin on all nodes.

This will incur downtime. Can you please help me understand how High
Availability can be achieved then.

Thanks in advance!





--
Sent from: http://apache-kylin.74782.x6.nabble.com/

Reply via email to