Indexer Nodes are under extrem Load

Hi there,

we created a Couchase Cluster with 9 Nodes:

3 Datanodes
3 Index Nodes
3 Query Nodes

We will run it on Ubuntu with in a LXC Container.

So all is running fine as expected. But Sometimes just the Index Nodes are raising the Load to 200/300 for couple of hours. System is stucking for a couple of hours.

Logs say nothing special.

I gave all index Nodes: 64GB Ram - 16 CPUs and 500GB Harddisk.

We just have one bucket with round about 2 MIO Items. Is there any Idea what can be faulty?




@daniel, what is the couchbase version and what is the type of indexes(Standard GSI or Memory optimized)?

You can take a look at the UI mini graphs for each index and see how much is the data size. And if the cumulative sum of all index size will fit in the memory you have on each node.

Another thing to look at would be how much is the query load during that peak utilization and if the queries are sub-optimal causing lot of full index scans.