Shrink issues through ILM

hi there,
I have some questions about using shrink through ILM. First, I set the index to shrink to 1 shard when it moves to the warm phase. I encountered two issues:

  1. When shards relocates (transfer to warm node), it only considers if there is enough space to hold all shards. When all shards are concentrated on the same node then shrink index is generated, but it cannot be allocated due to the disk reach the watermark limit (read-only mode). I believe Elastic should allocate to a node with at least twice the space needed when an index needs to shrink. Alternatively, do you have any other suggestions besides expanding my storage capacity?
  2. When an index moves from hot to warm and the warm node does not exist at that time, the index will be left on the hot node. Should the index stop shrinking at this point, instead of creating a shrink index that cannot be allocated on a hot node?