Use elastic create operation instead of index

[jeremyross]

create makes more sense and allows micrometer to write to elastic data streams.

Closes #2996

[pivotal-cla]

@jeremyross Please sign the Contributor License Agreement!

Click here to manually synchronize the status of this Pull Request.

See the FAQ for frequently asked questions.

[pivotal-cla]

@jeremyross Thank you for signing the Contributor License Agreement!

[shakuzen]

Is this a breaking change for users not using data streams?

[jeremyross]

No, it's completely transparent. I've tested this change with indexes and data streams. The index name specified can be an index or a data stream. And if no index or data stream exists, elastic will create one and the matching index template determines whether an index or data stream is created.

[shakuzen]

I don't know why the CI wasn't run for this pull request, but unit tests need to be updated for this change. They are failing now.

[patpatpat123]

Hello @jeremyross ,

Hope this message finds you well.
First of all, if not anything else, big thanks for this PR!

I took the latest builds of micrometer, from 1.10 all the way to the latest milestone.
I cross-tested with Elastic cluster from 7.6, all the way to the latest 8.7.

I am observing the following: those are facts and observations:

• test case 1: with this change, if the index already exists, and is of type plain index, micrometer will publish the metrics inside this plain index.
• test case 2: with this change, if the index already exists, and is of type data stream, micrometer will publish the metrics inside this data stream.
• test case 3: with this change, if the index does not already exist, meaning the index does not even exist in the first place, neither plain index, neither data stream, micrometer will publish the metrics, and a plain index will be created.

Above, for test cases 1, 2, and 3, by plain index, I mean this: (please see screenshot)

and by data stream I mean this (please see screenshot)

(small nota, your change is great for test case 2. Without your/this change, test case 2 would fail, I just verified)

Just wanted to confirm with you this is your expectation for the above test cases.

I am reaching out because for test 3, an expected enhancement/feature would be:

if the index does not already exist, meaning the index does not even exist in the first place, neither plain index, neither data stream, micrometer will publish the metrics, and a time series data stream will be created, that would be a great feature.

What do you think?

Finally, would you mind sharing the data stream index template for the micrometer metrics please?

Thank you for your time reading me.