Unequal load on nodes within single cluster on EC2

Elastic Stack Elasticsearch
1

Hi,

I'm having problems with load distribution between nodes within single
cluster. ES version is 0.90.12. My configuration is 5 nodes (m3.xlarge
machines on EC2) located in 3 different zones (2+2+1) with EC2 discovery.
All nodes with data. Index has 5 shards with 2 replicas, and is distributed
using force awareness with attribute "zone" set to "zone1", "zone2" or
"zone3". All nodes works well and process search requests. Cluster state is
green. I've got heap memory increased to 8GB.

Problem starts when I'm not using load balancer and all search requests
hits the same node. It looks line ES doesn't distribute work between all
nodes within the cluster, but only the ones within the same zone. Search
queue for these nodes is growing, while machines from different zones has
CPU usage 0% and no searches (I'm using bigdesk stats). During this test I
send only search requests, nothing else. Most parameters have default
values (nodes, index).

(When I use only single zone for all 5 nodes, something similar happens -
only set of nodes that together contains all shards is working, while the
rest of nodes do nothing)

Is there any reason why ES behaves in such a way?

Thanks!
Tom

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/6e764719-4b9d-4036-b0eb-8dbeb8d45657%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

2

Is it this?

Nik

On Thu, Apr 24, 2014 at 9:36 AM, Tomasz Danielczyk <
tomasz.danielczyk@gmail.com> wrote:

Hi,

I'm having problems with load distribution between nodes within single
cluster. ES version is 0.90.12. My configuration is 5 nodes (m3.xlarge
machines on EC2) located in 3 different zones (2+2+1) with EC2 discovery.
All nodes with data. Index has 5 shards with 2 replicas, and is distributed
using force awareness with attribute "zone" set to "zone1", "zone2" or
"zone3". All nodes works well and process search requests. Cluster state is
green. I've got heap memory increased to 8GB.

Problem starts when I'm not using load balancer and all search requests
hits the same node. It looks line ES doesn't distribute work between all
nodes within the cluster, but only the ones within the same zone. Search
queue for these nodes is growing, while machines from different zones has
CPU usage 0% and no searches (I'm using bigdesk stats). During this test I
send only search requests, nothing else. Most parameters have default
values (nodes, index).

(When I use only single zone for all 5 nodes, something similar happens -
only set of nodes that together contains all shards is working, while the
rest of nodes do nothing)

Is there any reason why ES behaves in such a way?

Thanks!
Tom

--
You received this message because you are subscribed to the Google Groups
"elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an
email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/elasticsearch/6e764719-4b9d-4036-b0eb-8dbeb8d45657%40googlegroups.comhttps://groups.google.com/d/msgid/elasticsearch/6e764719-4b9d-4036-b0eb-8dbeb8d45657%40googlegroups.com?utm_medium=email&utm_source=footer
.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/CAPmjWd2BvJ%2BTrtNP9m7wd9Dovb-ZGoZ6M3xzJEQtW2o4drB70A%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.