Elasticsearch 5.2.2 disk usage


(DanH) #1

We have a 3node ES cluster, and free disk space has become an issue over the weekend. We don't have the ability at the moment to add any more capacity, so we're looking at what we can remove.

We've audited our indices and set them to delete, and our expectation was that the disk apace would free up automatically, but it did not.

As a next step, we restarted ES on one of the nodes, and this also made no difference,

Lastly, we ran a forcemerge (per this article, and many like it), and this has also not made a difference.

Can anyone offer any guidance on how we can force a cleanup job of the deleted items, please?


(Mark Walkom) #2

How did you delete the indices?


(DanH) #3

Hi @warkolm,

We used DELETE in the API to remove the oldest/un-needed indices.


(Mark Walkom) #4

Once you remove them with the APIs they should no longer exist on disk at all. A force merge will not do anything, it's not delayed.

If you are still seeing disk issues, perhaps it is not the indices causing it?


(DanH) #5

We've put our collective heads together and found the root cause.

In the last few months, much more data has been going into the cluster, so we're not able to keep as much data as we did before.

Thanks very much for your help. :slight_smile:


(Mark Walkom) #6

You should install X-Pack and enable Monitoring. It's free and will give you good insights into your cluster.


(system) #7

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