Hi,

I started spark standalone cluster with 4 nodes each running around 8 core.
I want to scale it to even larger number of nodes.

The issue is that I am noticing lot of worker failure.

I believe I am looking for a zookeeper based coordination where if a worker
fails, another one is added from a pool of zookeeper host.

I am building from the github master. Can I use the zookeeper patch for
standalone cluster ?

https://github.com/apache/incubator-spark/pull/19

Thanks.
Deb

Reply via email to