We have a 9 node cluster, 3 masters, 6 data. We've been using the java
transport client, which connects to all 3 masters. Occassionaly the masters
become unresponsive and needs to restart.
Should the transport clients connect to the 6 data nodes directly instead?
We have a 9 node cluster, 3 masters, 6 data. We've been using the java
transport client, which connects to all 3 masters. Occassionaly the masters
become unresponsive and needs to restart.
Should the transport clients connect to the 6 data nodes directly instead?
We give the master nodes 5gb of memory -- stats are showing low cpu &
memory utilization. Would you still recommend the client only node? If so,
how many & powerful?
On Saturday, January 17, 2015 at 6:55:12 PM UTC-8, Mark Walkom wrote:
Depends, sounds like you need a few client nodes if you are OOMing your
masters (which, is a bad thing to happen to masters).
On 18 January 2015 at 10:23, Justin Zhu <haora...@gmail.com <javascript:>>
wrote:
We have a 9 node cluster, 3 masters, 6 data. We've been using the java
transport client, which connects to all 3 masters. Occassionaly the masters
become unresponsive and needs to restart.
Should the transport clients connect to the 6 data nodes directly instead?
We give the master nodes 5gb of memory -- stats are showing low cpu &
memory utilization. Would you still recommend the client only node? If so,
how many & powerful?
On Saturday, January 17, 2015 at 6:55:12 PM UTC-8, Mark Walkom wrote:
Depends, sounds like you need a few client nodes if you are OOMing your
masters (which, is a bad thing to happen to masters).
We have a 9 node cluster, 3 masters, 6 data. We've been using the java
transport client, which connects to all 3 masters. Occassionaly the masters
become unresponsive and needs to restart.
Should the transport clients connect to the 6 data nodes directly
instead?
Would all transport clients only connect to this client node? Right now we
have them connecting to all 3 master node.
On Sunday, January 18, 2015 at 8:43:08 PM UTC-8, Mark Walkom wrote:
It depends on your use, but try adding one client in with 8GB heap and see
how you go.
On 19 January 2015 at 16:48, Justin Zhu <haora...@gmail.com <javascript:>>
wrote:
We give the master nodes 5gb of memory -- stats are showing low cpu &
memory utilization. Would you still recommend the client only node? If so,
how many & powerful?
On Saturday, January 17, 2015 at 6:55:12 PM UTC-8, Mark Walkom wrote:
Depends, sounds like you need a few client nodes if you are OOMing your
masters (which, is a bad thing to happen to masters).
We have a 9 node cluster, 3 masters, 6 data. We've been using the java
transport client, which connects to all 3 masters. Occassionaly the masters
become unresponsive and needs to restart.
Should the transport clients connect to the 6 data nodes directly
instead?
We give the master nodes 5gb of memory -- stats are showing low cpu &
memory utilization. Would you still recommend the client only node? If so,
how many & powerful?
On Saturday, January 17, 2015 at 6:55:12 PM UTC-8, Mark Walkom wrote:
Depends, sounds like you need a few client nodes if you are OOMing your
masters (which, is a bad thing to happen to masters).
We have a 9 node cluster, 3 masters, 6 data. We've been using the java
transport client, which connects to all 3 masters. Occassionaly the masters
become unresponsive and needs to restart.
Should the transport clients connect to the 6 data nodes directly
instead?
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.