How to fix Unstable Clustering of elastic search!

(Brett Jang) #1

We are now using ES version 1.7.3
Recently we have founded one problem with this Ver.1.7.3 ,
The problem is about unstable clustering operations. This is not frequent failure but rarely happens especially when we shut down by CLI command while ES is collecting Data, after rebooting then cluster is broken. (OS is on Linux Ubuntu 12.04, HW is composed of one Master node and two data nodes,
This is not Hadoop DB but just locally distributed clusters)
This is tough explanation about the failure but please help me out with any tips and informations!
I don’t know why the reason is and don’t know how it can be fixed? Is this happened because of using still low E/S version?) Is there anybody who experienced same situation?
This can cause big problem sometimes for my customer.
Please tell me any helpful tips!

Hope answers!

(Jörg Prante) #2

You obviously selected to use only one master node.

So you are fully responsible for your cluster failures.

(Brett Jang) #3

Dear Jorg Prante!
Thank you your comments!
I'll check the problem and construct more master Node!
then let's talk again!
Thank you very much!

(Christian Dahlqvist) #4

Make you 2 data nodes master eligible or add another 2 dedicates master nodes. You basically always want to have 3 master eligible nodes so that the remaining 2 can form a majority of one goes down. Also remember to set minimum_master_nodes to 2 when you have 3 master eligible nodes in order to avoid split-brain scenarios.

(Brett Jang) #5

Dear Chritian_Dahlqvist,

Thanks for helping me out!

I’ll check that out!


(system) #6

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.