Index clone v.s. ILM

[nisow95612]

Hello nice people of Elasticsearch,

what is the proper way to apply ILM to cloned indices?
I know I can set index.lifecycle.origination_date so ILM knows how old data is,
but that often causes the index to go through multiple relocations and force-merges.

I assume this is because ILM treats the cloned index as if it was new.
Is there a way to clone the ILM state along with the index, so this does not happen?

Thank you!

[nisow95612]

Currently I am using this workaround:

• clone with index.lifecycle.origination_date set to something benign like date -d yesterday +%s000 and number_of_replicas set to 0 (as of 8.2, clone API can't clone replicas)
• wait for index status green
• read correct ILM state using Explain lifecycle API | Elasticsearch Guide [8.2] | Elastic
• force ILM to correct state using Move to lifecycle step API | Elasticsearch Guide [8.2] | Elastic
• set actual value of index.lifecycle.origination_date and number_of_replicas