Couldnt connect to host ElasticSearch down?

Elastic Stack Elasticsearch
1

Hi All,

I face a problem to insert data into ElasticsSearch sometimes and i catch the exception is "Couldnt connect to host[ip address] , ElasticSearch down?". I am using ElasticSearch version is "1.7.5". and i am using php ElasticSearch client

cluster detail is 1 master node and 3 data node.
Master node ( 4 cpu, 13G Ram, 8G used Ram )
Data node 1 ( 4 cpu, 27G Ram, 25G used Ram )
Data node 2 ( 4 cpu, 27G Ram, 26G used Ram )
Data node 3 ( 4 cpu, 27G Ram, 26G used Ram )

Please help me out from this problem.

2

Can you paste the exception message, the relevant initialization code of your PHP client, and the addresses of your nodes?

3

First of all thanks for reply.

$attribute=array("35411" => "2016-07-19T00:04:23", "35413" => "2016-07-19T00:04:23");

try {
$ElasticSearchConnection = \ElasticSearch\Client::connection(array( 'servers' => 127.0.0.1, 'protocol' => 'http', 'type' => 'typeName', 'index' => 'indexName'));
$ElasticSearchConnection->index($attribute, 1);
} catch (Exception $ex) {
print_r($ex->getMessage());
}

exception message is Couldnt connect to host [127.0.0.1], ElasticSearch down?

Please provide the best solution

4

Can you confirm Elasticsearch is running locally? What's the output of:

curl -XGET 127.0.0.1:9200

Maybe you need to specify the port in the 'servers' parameter? Perhaps open an issue with the author of your client and ask there?

5

Actually i am using cluster on production but sometime its through exception "Couldnt connect to host , ElasticSearch down?"

I get message from RabbitMq Queue and then insert it into ElasticSearch but 5 % document not insert into ElasticSearch.

All ports are open.

6

Do you have some logs from your production cluster? Can you reach the cluster via curl when the client is reporting failures?

The error sounds pretty explicit: it cannot connect to your cluster. So there's either a temporary network event which is preventing the client from connecting, or something on the cluster which is preventing connections (such as long-running GCs). Or potentially because the node it is trying to talk down has gone down for some reason.