We're currently operating a v2.4 cluster, and want to upgrade to v6.x
As I understand it, v6.x wont read v2.4-created indices, but will read v5.6.9-created indices.
So the plan is go to to v5.6.9, reindex all the data into v5.6.9 created-indices, and then upgrade to v6.x. That way we only have to reindex once, which is good, because we have a LOT of data. (Like, seriously, a LOT)
Bear with me.
In our v2.4 cluster we're using the _all
field. v5.6.9 supports _all
, but it's deprecated in v6.x, and the docs suggest using a copy_to
field.
My question is: what should my 5.6.9 mapping look like with regards to the _all
field to ensure that it's totally compatible with v6.x and later indices?
In my v6.x indices mappings I was going for something like this for an all field:
"all": {
"type": "text",
"store": false
}
with relevant fields having copy_to: all
. Is that a good way of doing it?
Will copy_to
in v5.6.9 work in the same way it does in v6.x?
Are people expecting that when v7.x arrives it won't be able to read v5.x indices?
Thanks!