We have a 2.44 cluster that is used for logging. To get perfomance and better bang for bug I'm trying to switch our architecture to a hot-warm setup. The hot nodes will run on the new AWS i3.xlarge nodes with attached SSD, warm on ebs backed volumes.
Since switching, I've seen our garbage collection frequency increase dramatically and stick to an increasingly narrow range of values that is well below max heap size. see marvel below
How can I get elasticsearch to use more of its HEAP and decrease the frequency of collections? Any help would be appreciated
Node details below (some data left out so it can fit in post)
"cluster_name": "mmelasticsearch",
"nodes": {
"7272HATlSFmRvsnp9ro9nw": {
"name": "Constance",
"attributes": {
"aws_availability_zone": "ap-southeast-2c",
"box_type": "hot",
"master": "1"
},
"index": {
"analysis": {
"filter": {
"filterShort2": {
"type": "length",
"max": "1000",
"min": "2"
}
},
"analyzer": {
"plusOne": {
"filter": [
"filterShort2"
],
"type": "custom",
"tokenizer": "tokenShort2"
}
},
"tokenizer": {
"tokenShort2": {
"type": "standard"
}
}
},
"number_of_shards": "2",
"number_of_replicas": "1"
},
"foreground": "false",
"threadpool": {
"bulk": {
"queue_size": "4000"
},
"search": {
"queue_size": "2000"
},
"get": {
"queue_size": "2000"
}
},
"node": {
"name": "Constance",
"box_type": "hot",
"master": "1"
},
"indices": {
"fielddata": {
"cache": {
"size": "40%"
}
}
},
"name": "Constance",
"client": {
"type": "node"
},
"config": {
"ignore_system_properties": "true"
}
},
"os": {
"refresh_interval_in_millis": 1000,
"name": "Linux",
"arch": "amd64",
"allocated_processors": 4
},
"process": {
"refresh_interval_in_millis": 1000,
"id": 17856,
"mlockall": false
},
"jvm": {
"pid": 17856,
"version": "1.8.0_121",
"vm_name": "OpenJDK 64-Bit Server VM",
"mem": {
"heap_init_in_bytes": 16106127360,
"heap_max_in_bytes": 16071262208,
"non_heap_init_in_bytes": 2555904,
"non_heap_max_in_bytes": 0,
"direct_max_in_bytes": 16071262208
},
"gc_collectors": [
"ParNew",
"ConcurrentMarkSweep"
],
"memory_pools": [
"Code Cache",
"Metaspace",
"Compressed Class Space",
"Par Eden Space",
"Par Survivor Space",
"CMS Old Gen"
],
"using_compressed_ordinary_object_pointers": "true"
},
"transport": {
"bound_address": [
"[::1]:9300",
"127.0.0.1:9300",
"10.28.140.49:9300"
],
"publish_address": "10.28.140.49:9300",
"profiles": { }
},
"http": {
"bound_address": [
"[::1]:9200",
"127.0.0.1:9200",
"10.28.140.49:9200"
],
"publish_address": "10.28.140.49:9200",
"max_content_length_in_bytes": 104857600
},
"plugins": [
{
"name": "cloud-aws",
"isolated": true,
"site": false
},
{
"name": "license",
"site": false
},
{
"name": "marvel-agent",
}
],
"modules": [
{
"name": "lang-expression",
},
{
"name": "lang-groovy",
},
{
"name": "reindex",
}
]
}
}