fix(spv): filter headers not starting from checkpoint on reconnect

[pauldelucia]

If you sync from checkpoint, disconnect, and reconnect, filter headers were being loaded from storage at 0 instead of checkpoint.

Summary by CodeRabbit

• Bug Fixes
  • Fixed blockchain sync height calculation during checkpoint-based synchronization so the next sync height never regresses when the node's tip is behind the configured sync base, improving synchronization correctness and stability.

_{✏️ Tip: You can customize this high-level summary in your review settings.}

[coderabbitai]

Walkthrough

The store_filter_headers function's next-blockchain-height calculation was adjusted: when a checkpoint sync base exists and the current tip is below that checkpoint, the next height is set to the checkpoint; otherwise the prior tip+1 behavior remains. The None-tip path is unchanged.

Changes

Cohort / File(s)	Change Summary
Height calculation refinement dash-spv/src/storage/disk/filters.rs	Modified next_blockchain_height calculation in store_filter_headers: for Some(tip) if sync_base_height > 0 and tip < sync_base_height then set next_blockchain_height = sync_base_height, otherwise set next_blockchain_height = tip + 1. The None case is unchanged.

Sequence Diagram(s)

sequenceDiagram
  participant Caller as store_filter_headers
  participant Storage as FiltersStorage

  Caller->>Storage: request store headers (has sync_base_height)
  alt tip is Some
    Storage->>Storage: if sync_base_height > 0 and tip < sync_base_height
    alt tip < sync_base_height
      Storage-->>Caller: next_blockchain_height = sync_base_height
    else tip >= sync_base_height
      Storage-->>Caller: next_blockchain_height = tip + 1
    end
  else tip is None
    Storage-->>Caller: next_blockchain_height = (unchanged) sync_base_height/path
  end
  Note right of Storage `#D3E4CD`: Decision focused on tip vs checkpoint\n(unchanged None-case)

Loading

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

• Verify the conditional correctly handles tip < sync_base_height.
• Confirm the None-case remains correct in surrounding logic.
• Check for off-by-one or integer underflow/overflow edge cases.

Poem

🐰 I nibble code where checkpoints lay,
If tips fall short, I hop and stay—
Keep heights firm until the climb,
No eager leap ahead of time.
A tiny tweak, a steadier pace, hooray!

Pre-merge checks and finishing touches

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and specifically describes the main fix: filter headers not resuming from checkpoint on reconnect, which matches the primary change in the PR.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.

✨ Finishing touches

• 📝 Generate docstrings

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment
• Commit unit tests in branch fix/filters-resync

---

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: 0

🧹 Nitpick comments (1)

dash-spv/src/storage/disk/filters.rs (1)

22-30: Checkpoint-aware next_blockchain_height logic correctly fixes the reconnect behavior

Anchoring next_blockchain_height to sync_base_height when sync_base_height > 0 and cached_filter_tip_height is still below the checkpoint ensures that new filter headers are written starting at the checkpoint, not into pre-checkpoint indices. That keeps the checkpoint-relative storage mapping consistent on reconnect and avoids mixing “genesis-style” indices with checkpoint-based ones, which aligns with the PR objective.

Given this guard, the next_blockchain_height < sync_base_height branch in the storage index computation (Lines 48–56) should now be unreachable in normal operation. You might consider tightening that path to make the invariant explicit, e.g. via a debug_assert!(next_blockchain_height >= sync_base_height) and keeping the tracing::warn! as a defensive fallback, so any future regressions are easier to spot during development.

Also applies to: 44-56

📜 Review details

Configuration used: CodeRabbit UI

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 4e9ef04 and 0b163c5.

📒 Files selected for processing (1)

• dash-spv/src/storage/disk/filters.rs (1 hunks)

🧰 Additional context used

🧠 Learnings (2)

📓 Common learnings

Learnt from: DCG-Claude
Repo: dashpay/rust-dashcore PR: 0
File: :0-0
Timestamp: 2025-06-26T16:01:37.609Z
Learning: The mempool tracking infrastructure (UnconfirmedTransaction, MempoolState, configuration, and mempool_filter.rs) is fully implemented and integrated in the Dash SPV client as of this PR, including client logic, FFI APIs, and tests.

📚 Learning: 2025-06-26T16:01:37.609Z

Learnt from: DCG-Claude
Repo: dashpay/rust-dashcore PR: 0
File: :0-0
Timestamp: 2025-06-26T16:01:37.609Z
Learning: The mempool tracking infrastructure (UnconfirmedTransaction, MempoolState, configuration, and mempool_filter.rs) is fully implemented and integrated in the Dash SPV client as of this PR, including client logic, FFI APIs, and tests.

Applied to files:

• dash-spv/src/storage/disk/filters.rs

[coderabbitai]

Actionable comments posted: 0

🧹 Nitpick comments (1)

dash-spv/src/storage/disk/filters.rs (1)

22-28: Checkpoint resume behavior looks correct; consider logging the unexpected tip < sync_base_height case.

The new conditional cleanly fixes the reconnect-from-checkpoint issue: when a cached tip is behind the checkpoint, you now restart from sync_base_height instead of tip + 1, while leaving genesis-sync and normal continuation (tip ≥ checkpoint) semantics unchanged. That’s consistent with how storage indices are derived later in the loop and avoids writing headers at pre‑checkpoint indices.

Since tip < sync_base_height should generally not happen in a healthy checkpointed flow (other than the specific reconnect scenario you’re targeting), you might optionally emit a tracing::warn! in that branch to surface any future unexpected state transitions:

Some(tip) => {
    if sync_base_height > 0 && tip < sync_base_height {
        tracing::warn!(
            "cached_filter_tip_height ({}) is below sync_base_height ({}); \
             resetting next_blockchain_height to checkpoint",
            tip,
            sync_base_height
        );
        sync_base_height
    } else {
        tip + 1
    }
}

This keeps the fix while making it easier to spot misconfigurations or future regressions around checkpoint handling.

📜 Review details

Configuration used: CodeRabbit UI

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 0b163c5 and 050a5cd.

📒 Files selected for processing (1)

• dash-spv/src/storage/disk/filters.rs (1 hunks)

🧰 Additional context used

🧠 Learnings (1)

📓 Common learnings

Learnt from: DCG-Claude
Repo: dashpay/rust-dashcore PR: 0
File: :0-0
Timestamp: 2025-06-26T16:01:37.609Z
Learning: The mempool tracking infrastructure (UnconfirmedTransaction, MempoolState, configuration, and mempool_filter.rs) is fully implemented and integrated in the Dash SPV client as of this PR, including client logic, FFI APIs, and tests.

[github-actions]

This PR has merge conflicts with the base branch. Please rebase or merge the base branch into your branch to resolve them.