influxdb-2.4: capture db.operation.batch.size for BatchPoints writes

[bhuvan-somisetty]

Resolves #19264

Description

influxdb-2.4 records db.operation.name for writes, but a write through InfluxDB.write(BatchPoints) sends a whole batch of points in a single request and the point count was never captured. This meant a batch write of, say, 500 points looked identical on the span to a write of a single point.

db.operation.batch.size is a stable database semantic-convention attribute, and the point count is readily available on the BatchPoints object the instrumentation already has in hand (BatchPoints.getPoints()). This follows the same pattern already used by the hbase-client instrumentation for its batch operations.

Changes

• Added a nullable batchSize field to InfluxDbOperation (the @AutoValue request object).
• In InfluxDbImplInstrumentation.InfluxDbModifyAdvice#onEnter, when arg0 instanceof BatchPoints, read ((BatchPoints) arg0).getPoints().size() and pass it into InfluxDbOperation.create(...). It stays null for the UDP / createDatabase / deleteDatabase paths, where there is no batch.
• Overrode getDbOperationBatchSize in InfluxDbAttributesGetter to return that value.
• Updated InfluxDbClientTest#testQueryAndModifyWithOneArgument to assert db.operation.batch.size = 2 on the write span for the existing two-point BatchPoints write.

No new configuration is needed. The shared DbClientAttributesExtractor already emits db.operation.batch.size under stable semconv and treats a size of 1 as a non-batch operation, so single-point writes are unaffected.

Test plan

• ./gradlew :instrumentation:influxdb-2.4:javaagent:compileJava :instrumentation:influxdb-2.4:javaagent:compileTestJava
• ./gradlew :instrumentation:influxdb-2.4:javaagent:spotlessCheck :instrumentation:influxdb-2.4:javaagent:checkstyleMain :instrumentation:influxdb-2.4:javaagent:checkstyleTest
• ./gradlew :instrumentation:influxdb-2.4:javaagent:test (requires Docker/testcontainers, run in CI)

[opentelemetry-pr-dashboard]

Pull request dashboard status

Status last refreshed: 2026-07-21 03:17:02 UTC.

• Waiting on: Reviewers
• Next step: Review the latest changes.

This automated status or its linked feedback items may be incorrect. If something looks wrong, report it with the result you expected.

[Copilot]

Pull request overview

Adds stable semantic-convention batch-size telemetry for InfluxDB BatchPoints writes.

Changes:

• Captures the number of points in batch writes.
• Exposes it through InfluxDbAttributesGetter.
• Verifies stable-semconv emission in the existing integration test.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated no comments.

File	Description
InfluxDbOperation.java	Stores the nullable batch size.
InfluxDbImplInstrumentation.java	Extracts the BatchPoints point count.
InfluxDbAttributesGetter.java	Supplies the batch size to database attribute extraction.
InfluxDbClientTest.java	Asserts db.operation.batch.size = 2.

[bhuvan-somisetty]

All checks are green now. @trask @laurit @jaydeluca @zeitlinger would any of you have some time to take a look? You've reviewed my last few PRs (#19154, #19088, #19078) so tagging you again, but happy to hear from anyone else too.

[trask]

hi @bhuvan-somisetty! I'm not sure that this should be represented a batch operation, which is related to why I sent open-telemetry/semantic-conventions#3811, what do you think? thanks

[bhuvan-somisetty]

@trask Good point, and that clarification in semantic-conventions#3811 makes it clear: a BatchPoints.write() call is really a single write operation that accepts multiple point operands, the same way MGET isn't a batch just because it takes multiple keys. It doesn't fit "two or more operations explicitly submitted as separate operations."

I've reverted the db.operation.batch.size change in this PR back to what it was before. Thanks for flagging it!