Garbage Collection problem


(張皓翔) #1

Hi, my cluster has 4 nodes with 3 shards and 1 replica. Every node has 16G RAM and 8 cores CPU. About 50-60G data per day and now there's 55 indices, 234 shards and 2T total data.
The GC problem is a little serious, and how could I improve it?


Besides, I found that when I query through Dev Tools, it would return bad gateway and timeout sometimes and the index rate become very slowly. In addition, when I display the dashboard, the index rate would become very slowly too. How can I improve this problem?

thank you in advance :slight_smile:


(David Pilato) #2

May be 234 shards for 4 nodes is too much?


(張皓翔) #3

so 4 nodes with 2 primary shards is better?
If yes, I will tune it.

whether 2 primary shards slow down the query speed?


(David Pilato) #4

I think I already shared that with but let's do it in case I did not.

Look at the following resources about sizing:

https://www.elastic.co/elasticon/conf/2016/sf/quantitative-cluster-sizing


(張皓翔) #5

could the Force Merge improve the search query speed or reduce the nodes pressure?


(David Pilato) #6

Improve search: probably. As long as you are not writing anymore to the indices.
Reduce pressure: I'm unsure


(system) #7

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.