Here is the solution: https://stackoverflow.com/a/64010705/518704
Your docs are slightly malformed. If you want proper nested-ness, you'll need to refactor them like so:
{
"dtmSent":[
{
"dtmSent":"1990-01-01"
},
{
"dtmSent":"2000-05-01"
},
{
"dtmSent":"2020-01-01"
}
]
}
instead of the array-of-strings.
The index mapping can stay the same, the query too as well as the sorting part. After that, the order will work as you expect 1 -> 3 -> 2.
After you drop the index, here's the _bulk command to get you started (only after setting the mapping of course):
POST _bulk
{"index":{"_index":"my-index-000001","_type":"_doc","_id":1}}
{"dtmSent":[{"dtmSent":"1990-01-01"},{"dtmSent":"2000-01-01"},{"dtmSent":"2000-01-02"},{"dtmSent":"2020-01-01"}]}
{"index":{"_index":"my-index-000001","_type":"_doc","_id":2}}
{"dtmSent":[{"dtmSent":"1990-01-01"},{"dtmSent":"2000-05-01"},{"dtmSent":"2020-01-01"}]}
{"index":{"_index":"my-index-000001","_type":"_doc","_id":3}}
{"dtmSent":[{"dtmSent":"1990-01-01"},{"dtmSent":"2000-03-01"},{"dtmSent":"2020-01-01"}]}
{"index":{"_index":"my-index-000001","_type":"_doc","_id":4}}
{"dtmSent":[{"dtmSent":"1990-01-01"},{"dtmSent":"2002-03-01"},{"dtmSent":"2020-01-01"}]}