To update this, I've just found I can resolve the problem by stopping just
the master node. Instantly the other data nodes settle down into single
figure CPU usage and <1 load.
The layout is:
caesium:
node.master: true
node.data: false
sodium, potassium, francium:
node.master: false
node.data
Hi,,
The only things I see is ES_HEAP_SIZE with 50% this can bet set to 8gb.
I'm seeing some swap usage, you could disable swap completely.
hth,,
A.
On Monday, January 5, 2015 12:48:53 PM UTC+1, Darren Mansell wrote:
>
> Hi all,
>
> We have a 4 node VM dev cluster with 3 data nodes and 1 maste
