Windows 2008 ES Cluster Issue


(Katrina K) #1

Hi,

We have four Servers in Our ES Cluster. In one of the server we saw a
memory strange situation. here is my server detailed Specification

ES Version: 16.0

Cluster Details
No of Data Nodes : 3 (Node1,2&3)
No of No Data Noode: 1 (Node 4)

Index Details
number_of_shards : 6
number_of_replicas : 1

Server Details:

  1. Node 1, 3 & 4 installed OS is Windows 2003 Server Enterprise
    edition R2 64 bit SP2.
  2. Node 2 installed OS is Windows 2008 Server Enterprise edition 64
    bit SP2

Node 1 & 2 Configuration
Pysical RAM - 48 GB
ES Allocated RAM - 20 GB

Node 3 Configuration
Pysical RAM - 24 GB
ES Allocated RAM - 20 GB

Node 4 Configuration
Pysical RAM - 24 GB
ES Allocated RAM - 10 GB

Congig.yml Settings:
"network :
host : 10.41.23.4
gateway:
type: fs
fs:
location: \10.41.23.6\ElasticSearchGateway
indices:
memory:
index_buffer_size : 20%
index:
gateway:
snapshot_on_close: false
cluster:
name: Dragon
http:
enabled: false
index:
number_of_shards : 6
number_of_replicas : 1
bootstrap:
mlockall : true"

Memory Settings:
"if "%ES_MIN_MEM%" == "" (
set ES_MIN_MEM=20g
)

if "%ES_MAX_MEM%" == "" (
set ES_MAX_MEM=20g
)"

Node 2 Server(windows 2008) Elastic Search memory is increasing(full
usage) frequently compare to other servers (Windows 2003 Server). so
the cluster performance geeting slow for every time.

please help me how to avoid this issue
Is there any configuration item I'm missing?

Thanks,
Katrinaa


(Shay Banon) #2

First, 0.16.2 has much better memory usage patterns compared to 0.16.0. So, I suggest you give it a go.

One more thing, your installation is a bit problematic since the Node3 allocates 20gb out of 24gb to elasticsearch, which leaves very little to the operating system (and the file system cache).

On Saturday, June 11, 2011 at 1:20 AM, Katrina K wrote:

Hi,

We have four Servers in Our ES Cluster. In one of the server we saw a
memory strange situation. here is my server detailed Specification

ES Version: 16.0

Cluster Details
No of Data Nodes : 3 (Node1,2&3)
No of No Data Noode: 1 (Node 4)

Index Details
number_of_shards : 6
number_of_replicas : 1

Server Details:

  1. Node 1, 3 & 4 installed OS is Windows 2003 Server Enterprise
    edition R2 64 bit SP2.
  2. Node 2 installed OS is Windows 2008 Server Enterprise edition 64
    bit SP2

Node 1 & 2 Configuration
Pysical RAM - 48 GB
ES Allocated RAM - 20 GB

Node 3 Configuration
Pysical RAM - 24 GB
ES Allocated RAM - 20 GB

Node 4 Configuration
Pysical RAM - 24 GB
ES Allocated RAM - 10 GB

Congig.yml Settings:
"network :
host : 10.41.23.4
gateway:
type: fs
fs:
location: \10.41.23.6\ElasticSearchGateway
indices:
memory:
index_buffer_size : 20%
index:
gateway:
snapshot_on_close: false
cluster:
name: Dragon
http:
enabled: false
index:
number_of_shards : 6
number_of_replicas : 1
bootstrap:
mlockall : true"

Memory Settings:
"if "%ES_MIN_MEM%" == "" (
set ES_MIN_MEM=20g
)

if "%ES_MAX_MEM%" == "" (
set ES_MAX_MEM=20g
)"

Node 2 Server(windows 2008) Elastic Search memory is increasing(full
usage) frequently compare to other servers (Windows 2003 Server). so
the cluster performance geeting slow for every time.

please help me how to avoid this issue
Is there any configuration item I'm missing?

Thanks,
Katrinaa


(system) #3