I am hitting my 6 node cluster using a client node. I am running few aggregation queries as well as search queries.
I am getting out of memory issue on client node only rest of the data and master working fine.
Are loading _source for search queries can cause this problem. I am fetching 5000 documents for some search requests. Does these document get loaded in memory which cause high heap usage ?
Looking for suggestions.
There's very little to go on here.
What version are you on? How big is your heap? What is the actual error? How large are your documents? What does the search look like?
Here are the answers to you questions.
Version - 2.3.0
Heap Allocated to Client is 8 GB
Error - java.lang.OutOfMemoryError: Java heap space
Document size ~ 5 KB
Search - It is single range filter with sort on single field.
I really like here to validate that document from search query get loaded in memory on client node ? and can cause load on memory ??
Provide the entire thing please.