High disk.used with low disk.indices

I am trying to figure out what is consuming disk space on my hot data node. When I run GET _cat/allocation, the output shows that disk.indices is only consuming 80.4gb, whereas disk.used is 648.4gb. On my other hot data node the discrepancy between these two metrics is small. Documentation says that disk.indices doesn't include unassigned shards or translogs, so how do I know what is taking up the 568gb of unaccounted-for space?

It could be due to unmerged segments, but that's a lot of space.

What does a df -h show on that node?

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