We do not use Scroll Query or PIT, What caused our ElasticSearch Cluster big OpenContexts Spike

Elastic Stack Elasticsearch
1

We experienced OpenContexts spikes in our Elasticsearch 6.8 environment. Since we’re not using Scroll Queries or Point in Time (PIT), I’d like to understand:

  • What could be causing these OpenContexts spikes?

Note: We have already ruled out Scroll Queries and PIT as potential causes.

2

Are you using update-by-query and/or delete-by-query by any chance? I believe these use the scan/scroll/PIT mechanism behind the scenes.

Also note that Elasticsearch 6.8 is very old and has been EOL a looong time, so I would recommend upgrading to the latest version if possible.

3

We are not using either Delete-by-Query or Update-by-Query.
Yeah it is part of our road map to update it but we expect to continue using Elasticsearch 6.8 for some years.

4

You can use what you want. It’s your system. What you cannot really expect is for the community to be hugely motivated to help you, unless it’s a fairly trivial query. 6.8 was released in 2019 or so? EOL has a meaning.

1 Like
5

Can’t think of anything else that might cause this right now apart from reindexing, but have not used Elasticsearch 6.8 in many years. It may help if you could test this on a recent and supported version to check if this is still an issue or perhaps provide more details about your use case and queries.

1 Like