fix: tolerate incomplete query traces in ShardAwarenessTest

[nikagra]

correctShardInTracingTest fails on ~27% of java-driver-matrix-test builds (master apache#2083, apache#2085, apache#2091; 2026.3 #4), always with:

No 'querying locally' trace event was observed for the query

QueryTrace.doFetchTrace() retries only until the coordinator's session row reports a duration, then freezes whatever the concurrently-issued events query returned — its own comment notes the trace "may not contain the log of replicas". So getEvents() can return an empty list, which the test read as "the coordinator never queried locally".

Retry the traced read up to 3 times, binding a fresh statement per attempt (reusing one would pin the coordinator via the paging optimization), and fail only when no attempt observes a local read. Re-executing is required: once fetched, a trace's events are cached. A wrong shard still fails immediately, without retrying.

Verified against 3-node Scylla 2026.1: 10/10 runs pass, one of which hit the real fault and recovered —

Attempt 1/3 for pk=100002: trace fa0a9170-… carried no 'querying locally'
    event (0 events); retrying with a fresh trace

pk=100002 is the shard-1 key that fails in CI. Sabotaging the matcher so nothing can match still yields BUILD FAILURE after 3 attempts.

🤖 Generated with Claude Code

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: QUIET

Plan: Pro Plus

Run ID: 9e9a30e2-6f9f-43bc-beec-be50f8a5816f

📥 Commits

Reviewing files that changed from the base of the PR and between 855dc8c and 428cc25.

📒 Files selected for processing (1)

• driver-core/src/test/java/com/datastax/driver/core/ShardAwarenessTest.java

🔗 Linked repositories identified

CodeRabbit considers these linked repositories for cross-repo context during reviews:

• scylladb/github-automation (auto-detected)
• scylladb/scylladb (auto-detected)

---

📝 Walkthrough

Walkthrough

ShardAwarenessTest now retries traced shard verification up to three times. Each attempt uses a fresh bound statement, validates the returned row, and checks for a data-local read on the expected shard. The test ignores unrelated coordinator events, normalizes service-level thread suffixes, logs incomplete traces, and fails explicitly when no valid local-read event appears.

Possibly related issues

• 3.x: ShardAwarenessTest.correctShardInTracingTest flakes when tracing lacks querying locally event #946 — Describes the flaky tracing verification addressed by bounded retries.

Possibly related PRs

• scylladb/java-driver#931 — Updates the same traced shard verification in ShardAwarenessTest.

Mergeability Score: ⚪ Minimal · up to 428cc

This PR makes the shard-awareness test retry incomplete traces while preserving immediate failure for wrong-shard results. No actionable merge-blocking risk remains beyond normal checks and review.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly and concisely describes the fix for incomplete query traces in ShardAwarenessTest.
Description check	✅ Passed	The description directly explains the intermittent failure, retry strategy, fresh statements, and validation behavior.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

---

_{Comment @coderabbitai help to get the list of available commands.}