Cluster doesn't respect routing.allocation.require.data_type

Hi,

we have cluster with 2 nodes. Every node has set data_type attribute.
One node has low disk watermark exceed.
We prepared ILM policy to keep indices on one node for some time and after move it to another.
Rarely I move some indices manually to another node by set index routing.allocation.require.data_type setting.

Few days ago one of node filled up disk because moving indices to another node stops working. When I manually set routing.allocation.require.data_type to value of second node data_type parameter nothing happens (I can't see any relocating shards in _cat/shards results. cluster/allocation/explain also doesn't show anything).

Few months ago I move indices this way without problems.
All (production) indices has 0 replicas so low disk watermark should not be a problem.

Have you any idea why indices doesn't move to another node?

Elastic version: 6.8.13

Thank you.

Cluster allocation explain is the best way to explain why a particular shard is allocated as it is. What do you mean it doesn't show anything? It will always say something about the target shard.

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