Hey!
We’re trying to manage default settings for a fairly large migration via snapshots from our old ES cluster.
Ideally we’d like the old indexes to be stored on our cold nodes, as well as create a default policy for all indexes which puts them in cold storage unless explicitly specified via SMP.
New indexes / incoming data will have a hot/cold management pipeline which is pretty standard.
My question:
- How do the allocation settings work when both are specified? Does the index setting take precedence?
cluster.routing.allocation.include.<attribute>
index.routing.allocation.include.<attribute>
-
If we set the allocation via lifecycle actions, does this overwrite the index
template? i.e. the template specifies hot, but the lifecycle has an action after 30 days that moves the index to a cold node. EDIT: Can confirm that lifecycle overwrites templates. Templates appear to only be set once upon index initialization. -
I’ve figured out how to force snapshots to be restored onto cold nodes by adding
index.routing.allocation.include.<attribute>
to the index settings when restoring, but I can’t seem to apply index templates or lifecycle policies to a snapshot automatically based on index patterns. Is this just impossible, or is there a way to force snapshots to use a template / policy?
I’d love to hear any thoughts on whether there’s a better way to manage this.
Thanks for the help!
- List item