Cluster runs out of memory and service stops during re-index

I use elasticsearch as our document content engine with HPE Content Manager. When I reindex my dataset the elasticservice stops due to bulkloader transactions timing out with the service. The memory reaches its threshhold. This has been happening on a physical server with 16GB of ram. I tried a trial of elastic service (cloud) and had the same probably with 30GB of ram.
My question is how can I reduce the memory pressure in my cluster? Can this be done by altering the mapping of my index?

I also have this issue so am interested in any answer.

In my case I have pinpointed a single document which it is trying to index. (The URI of the record in trouble can become clear if you watch the re-index process) I am doing a full re-index bit by bit and so far only the single document has caused this issue. I am nearly finished. Therefore I conclude that the structure of the document content is somehow problematical to Elasticsearch. It's a spreadsheet containing many rows of unique data but it's only 63Kb.

Welcome :slight_smile:

Can you share any logs or errors?
What version of Elasticsearch?
How are you starting the reindex?


It'd be better if you could create your own thread for this question :slight_smile:

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