Raft Cluster: fix CLUSTER SHARDS visibility and replica health reporting

[zuiderkwast]

Some fixes for the raft cluster bus:

1. Add nodes to shards dict on NODE_JOIN apply — Nodes joining the cluster were added to the nodes dict but not the shards dict, making them invisible in CLUSTER SHARDS.
   This affected both paths: fresh node creation on followers and MEET-flagged nodes transitioning to full members.

2. Broadcast repl_offset when it drops to zero — The leader broadcasts a follower's replication offset to peers when it changes from 0 to non-zero (replica finishes sync).
   Now also broadcasts when it drops from non-zero to 0 (replica starts full resync), so other nodes correctly report the replica's health as "loading" in CLUSTER SHARDS.

3. Include zero offsets in periodic REPL_OFFSETS broadcast — The periodic broadcast skipped nodes with offset 0, so a peer that missed the immediate transition broadcast (e.g. due to a brief disconnection) would never learn the correct value. Also fixes the case where the raft leader is a data replica and its offset drops to 0.

4. Defer first REPL_OFFSETS broadcast — Initialize the broadcast timer to startup time so the first periodic broadcast is deferred by 10 seconds, avoiding unnecessary
   traffic during cluster formation.

I discovered these when working on #3887, running the cluster-shards.tcl test suite, involving node restarts. These two fixes are not depending on persistence though.

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro Plus

Run ID: d8bc844f-7400-439d-adcb-c462897b3739

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 50.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title directly describes the two main fixes in the changeset: improving CLUSTER SHARDS visibility and replica health reporting in the raft cluster implementation.
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.
Description check	✅ Passed	The pull request description is directly related to the changeset, providing clear context about raft cluster fixes for CLUSTER SHARDS visibility and replica health reporting.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

---

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 and usage tips.}

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@src/cluster_raft.c`:
- Around line 1315-1318: The follower transition check only broadcasts when
prev_offset != follower_repl_offset and one side is zero, but the leader's own
path still only emits on 0->non-zero and periodic broadcasts skip zero offsets;
update the leader emission logic to also emit when the leader's replication
offset transitions from non-zero to 0 and ensure periodic broadcasts include
zero offsets so peers are updated. Concretely, modify the code paths that use
prev_offset and follower_repl_offset (and respect node->flags /
CLUSTER_NODE_MEET) to treat both 0->non-zero and non-zero->0 as broadcast-worthy
transitions and adjust the periodic broadcast logic to not omit zero offsets.
Ensure the same conditional symmetry used for followers is applied to the
leader-side emission.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 97ce5d13-d85d-453b-95f0-ade46b707bdb

📥 Commits

Reviewing files that changed from the base of the PR and between e67cc9c and 8091baf.

📒 Files selected for processing (1)

• src/cluster_raft.c

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 76.49%. Comparing base (e67cc9c) to head (d75a8fc).

Additional details and impacted files

@@              Coverage Diff               @@
##           cluster-v2    #3910      +/-   ##
==============================================
+ Coverage       76.47%   76.49%   +0.01%     
==============================================
  Files             166      166              
  Lines           82602    82605       +3     
==============================================
+ Hits            63174    63189      +15     
+ Misses          19428    19416      -12     

Files with missing lines	Coverage Δ
src/cluster_raft.c	61.26% <100.00%> (-0.99%)	⬇️

... and 21 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[murphyjacob4]

Makes sense to me