How to completely remove indices from disks after shard filtering definition

Hi, I wanted to create a simple two-node hot-warm cluster for quick experimentation. One ("warm") with already 100+GB of sample data and the other ("hot") with zero data initially. But I forgot to set shard filtering for some of the indices so they also exist on the hot node. So I corrected it and it seemed to have been moved to the right node. However, when I check the disk usage, it still uses the same amount of disk space (about 1 gigabyte) as before even though I told the shards to go to the "warm" node. If I delete the data folder and re-run the "hot" node, the size becomes only a few hundred kilobytes as expected. Does anybody know what the problem is here or if there is an API to claim the disk space?

I'm using Elasticsearch 2.3.1.


Side note - is there a reason you are using a version that is EOL?

Hi @warkolm, our production cluster is running on 2.3.1 so I'm using it locally as well to match.

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