fix: reset stale TTL on reused FlushRecord and bump rocksdb-cloud roll guard

[thweetkomputer]

Two changes:

• Bump third_party/src/rocksdb-cloud to 7b4210e (eloqdata/rocksdb-cloud main), which publishes the smallest-file-number guard object in RollNewCookie before the new epoch becomes reachable (Upgrade github workflow DoozyX/clang-format-lint-action to v0.18 #20) and fixes the const-qualification compile error in PutSmallestFileNumberObject ([Feature]: Move table stats broadcast into data sync worker #21).
• Fix silent data loss on RocksDB(-Cloud) backends: FlushRecord::SetNonVersionedPayload reuses its BlobTxRecord across data-sync scan batches but only cleared value_, leaving the previous record's ttl_ behind. A no-TTL object flushed into a contaminated slot was written to the KV store with another key's expiration, so the read path and the TTL compaction filter treated it as expired and the key disappeared once evicted from memory — no restart required. Reset ttl_ to UINT64_MAX (the "no TTL" sentinel) on every reuse.

🤖 Generated with Claude Code

Summary by CodeRabbit

• Bug Fixes

  • Corrected expiration handling when records without a time-to-live are reused, preventing stale expiration settings from affecting data.

• Chores

  • Updated the underlying storage infrastructure to a newer revision.
  • Incorporated upstream maintenance and improvements while preserving existing functionality.

[coderabbitai]

Walkthrough

The PR updates the rocksdb-cloud submodule reference and resets stale TTL metadata when BlobTxRecord payloads are reused without versioning.

Changes

RocksDB Cloud update

Layer / File(s)	Summary
Update submodule pointer third_party/src/rocksdb-cloud	The submodule reference advances to commit 7b4210e7351a961a7fc2196c0ecfc81d01be3ac8.

Blob TTL reset

Layer / File(s)	Summary
Reset reused blob TTL tx_service/include/cc/cc_entry.h	SetNonVersionedPayload resets the reused BlobTxRecord TTL to UINT64_MAX after clearing the previous value.

Estimated code review effort: 2 (Simple) | ~10 minutes

Possibly related PRs

• eloqdata/tx_service#547: Both PRs update the rocksdb-cloud submodule reference.
• eloqdata/tx_service#548: This PR updates the same submodule area, while its file-number guard changes are separate from the TTL reset.

Suggested reviewers: liunyl

Poem

A rabbit resets a stale TTL,
Then hops where cloud commits dwell.
Old payloads leave no trace behind,
A newer pointer leads the bind.

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Description check	⚠️ Warning	The description explains the changes and behavior impact but omits most required template sections, including tests, risks, rollback, reviewer guidance, and follow-up work.	Add the required sections with test commands and results, risk and rollback details, reviewer guidance, and explicit follow-up status.

✅ Passed checks (4 passed)

Check name	Status	Explanation
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.
Title check	✅ Passed	The title clearly summarizes both changes: the stale TTL fix and the rocksdb-cloud submodule update.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch update-rocksdb-cloud-roll-guard

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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