OpenSearch Transport v3.0 Implementation

[mznet]

This PR adds tracing for OpenSearch Transport interface, enabling in-depth tracing of OpenSearch requests across all Transport implementations.

Closes #14652

[jaydeluca]

Suggested change

	public class OpenSearchJavaApacheHttpClient5TransportTest extends AbstractOpenSearchJavaTest {
	class OpenSearchJavaApacheHttpClient5TransportTest extends AbstractOpenSearchJavaTest {

[jaydeluca]

Suggested change

	public class OpenSearchJavaAwsSdk2TransportTest extends AbstractOpenSearchJavaTest {
	class OpenSearchJavaAwsSdk2TransportTest extends AbstractOpenSearchJavaTest {

[jaydeluca]

Suggested change

	public class OpenSearchJavaRestClientTransportTest extends AbstractOpenSearchJavaTest {
	class OpenSearchJavaRestClientTransportTest extends AbstractOpenSearchJavaTest {

[SylvainJuge]

Because you are adding a new instrumentation, I think it would be a good idea to make it already "indy-ready" to make it future-proof and prevent having to rework it later with a PR like #14842.

I haven't looked closely, but this should be quite close to the equivalent instrumentation for Elasticsearch, which has already been made "indy-compliant".

[mznet]

@SylvainJuge good point, i haven't thought about that, i will look at "indy-ready"

[laurit]

Because you are adding a new instrumentation, I think it would be a good idea to make it already "indy-ready" to make it future-proof and prevent having to rework it later with a PR like #14842.

I haven't looked closely, but this should be quite close to the equivalent instrumentation for Elasticsearch, which has already been made "indy-compliant".

@mznet you can view this as an optional task, we'll merge this PR even if you don't do this

[laurit]

This probably isn't correct, ideally this should return the full query that is sanitized.

[mznet]

@laurit when i first wrote the implementation, extracting the full query text seemed impossible because the request only provided limited information. on second thought, it might actually be possible, so i’ll take another look.

[laurit]

Using the full query isn't always desirable either because it needs to be sanitized. If it can't be sanitized then it can't be enabled by default. Unfortunately sanitizing might not be easy. I think that since the existing opensearch-rest instrumentation uses the same query text we can use it here too for now. In the future we could use the same approach to fill the query summary (https://opentelemetry.io/docs/specs/semconv/database/database-spans/#generating-a-summary-of-the-query) instead and set the query text to null. Before we can do that we have to implement support for the db.query.summary attribute. Also since db.query.summary is only in the stable semconv would need to figure out what to do with the old semconv (continue filling the query text as is or set it to null). cc @trask

[mznet]

@laurit I was finally able to extract the full query text, but composing the sanitization logic will take more time. The existing sanitization logic can’t be reused since this isn’t just sql or sql-like but it’s a json query. I’ve roughly figured out how to approach opensearch query sanitization, but I’ll need more time to validate that it’s the right solution.

So I agree with your review. Let’s leave it as is for now, and I’ll come back once I’ve implemented the sanitization logic and can extract the full query text with it.

[mznet]

@SylvainJuge Since the review has already progressed this far, it might not be a good idea to change the code at this point. It would be better to add the indy-compatible instrumentation code in a separate PR. Thank you for pointing out. cc @laurit

[linux-foundation-easycla]

The committers listed above are authorized under a signed CLA.

• ✅ login: laurit / name: Lauri Tulmin (99b511b, c310326)
• ✅ login: mznet / name: Minje Park (1ec0416, 4443006, 4ed0def, 6451368, 8ea5cb4, a7663d1, ab8ca71, bde4f45, e4d95fa, e66488d, e90f611)

[mznet]

Looks like I did something wrong on this branch. I’ll redo it.

[mznet]

I did the rebase wrong, so other people’s commits ended up in this branch’s history. I force pushed to the pre-rebase commit and cherry-picked just the commits I wanted on top.

[laurit]

/easycla

[otelbot]

Thank you for your contribution @mznet! 🎉 We would like to hear from you about your experience contributing to OpenTelemetry by taking a few minutes to fill out this survey.