Set indices.fielddata.cache.size without restarting node?

Elastic Stack Elasticsearch
#1

Hi there guys. I'm trying to set my indices.fielddata.cache.size to 75% (with a circuit breaker of 60%, so we get alerts before we're close to out of memory), and I just had a couple of questions.

First, is that number 75% of the max memory on the machine, or is that 75% of ES_HEAP_SIZE? I figure its the latter, but I want to make sure I have a clear understanding.

Secondly, I have applied this setting in my elasticsearch.yml file, so it will take effect on start up, and when I build new nodes, however is there an api call to alter that setting on live nodes?? I'd prefer not to restart all of our data nodes if I do not have to in order to turn on this setting.

Thanks!

#2

Any limits you can set in ES are for the heap only, that is the only memory it addresses.

You should also read https://www.elastic.co/guide/en/elasticsearch/reference/2.3/circuit-breaker.html#circuit-breaker

#3

Thank you!

#4

Hmm I just added the setting using:

curl -XPUT "http://$(hostname --ip-address):9200/_cluster/settings" -d'{ "persistent" : { "indices.fielddata.cache.size" : "75%" } }'

and I received:

{"acknowledged":true,"persistent":{},"transient":{}}

in response, but when I run:

curl "http://$(hostname --ip-address):9200/_cluster/settings?pretty"

to look at my settings I just get back:

{
  "persistent" : {
    "discovery" : {
      "zen" : {
        "minimum_master_nodes" : "2"
      }
    },
    "cluster" : {
      "routing" : {
        "allocation" : {
          "enable" : "all",
          "cluster_concurrent_rebalance" : "15",
          "node_concurrent_recoveries" : "15"
        }
      }
    }
  },
  "transient" : {
    "cluster" : {
      "routing" : {
        "allocation" : {
          "enable" : "all"
        }
      }
    }
  }
}

I.e., I do not see it displaying the new setting I just enacted. Is this expected, or should I see it??

#5