Thread pool queue size is not updating in elasticsearch 5.x

[murman]

Hi,

I'm setting up elasticsearch 5.4.2 version, i'm trying to configure thread pool queue size in elasticsearch.yml file. I'm using below properties in elasticsearch.yml file:

thread_pool.bulk.queue_size: 5000

After restarting all nodes, if I validate in all nodes - this value is not updated.

GET _cluster/settings

This is returning as empty response:

{
   "persistent": {},
   "transient": {}
}

Can someone please help what I'm doing wrong here or am I missing anything?

Appreciate your help in advance!

Thanks!

[danielmitterdorfer]

Hi @murman,

the cluster settings API shows only settings that you've changed via the API itself, not settings from elasticsearch.yml. You can use the nodes info API though:

curl http://YOUR_HOST:9200/_nodes/thread_pool\?pretty

should show the relevant info. Here is the relevant snippet:

        "bulk" : {
          "type" : "fixed",
          "min" : 8,
          "max" : 8,
          "queue_size" : 5000
        }

I see that you want to increase the length of the bulk queue. Note that increasing the queue length is seldom the solution if the system is overwhelmed. You should either increase the system's capacity or throw less load at it.

Just think of a coffee shop with just one barista. If you, as the owner, see a constantly large queue of customers you should not think: Hey, let's move to a building with a longer hallway so all the customers will fit in my shop. You should rather hire a second barista because customers don't want to wait for hours to get one coffee. I hope the analogy makes my point a bit clearer.

Daniel