Problem
Say you have an Amazon OpenSearch domain in the red state due to UNASSIGNED shards. The indices in the red state had 2 primaries and no replicas, with one of the two primary shards UNASSIGNED. Backups are not available to restore all indices in the red state.
What I wanted to do
Now, in a self-managed cluster, I would minimise data loss using the _cluster/reroute API (allocate_empty_primary command) to reroute the UNASSIGNED primaries to another cluster node, retaining at least part of the documents in the index.
Why I cannot do it
However, the _cluster/reroute API is not supported in Amazon OpenSearch (https://docs.aws.amazon.com/opensearch-service/latest/developerguide/supported-operations.html), and the only option I'm aware of to recover the cluster from red state is to delete all red indices, also losing the data in the assigned primaries.
My question
Can you recommend another approach that could reduce data loss?
