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 `@node/cmd/node/main.go`:
- Around line 122-129: The Syncer has unsynchronized concurrent reads/writes to
the uint64 field latestSynced (accessed by fetchL1Messages(), the background
goroutine started in Start(), and readers via LatestSynced()), causing a data
race; fix by protecting all accesses to latestSynced: either add a sync.Mutex
(e.g., s.mu) and wrap every read/write in fetchL1Messages() and LatestSynced()
with s.mu.Lock()/Unlock(), or convert latestSynced to a uint64 and use
sync/atomic functions (atomic.LoadUint64/atomic.StoreUint64) in
fetchL1Messages(), Start() and LatestSynced() so all reads/writes are atomic and
race-free. Ensure you update any logging that reads latestSynced to use the
protected/atomic read.

---

Outside diff comments:
In `@node/derivation/metrics.go`:
- Around line 20-42: The file's formatting is out of sync with gofmt; run gofmt
on the Metrics struct file (e.g., the Metrics type and its fields like
L1SyncHeight, PathBTriggered, SafeAdvanceTotal, etc.) to normalize spacing and
alignment, then re-add the formatted file to your commit; you can run `gofmt -w
node/derivation/metrics.go` (or `gofmt -w .` at repo root) and re-run CI.

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 `@node/db/store.go`:
- Around line 214-217: The loop that calls s.db.Has(key) currently treats any
Has error as "key missing" and breaks, which swallows DB errors; change the
logic in the reorg cleanup loop so that if s.db.Has(key) returns a non-nil err
you raise/panic with that error (preserving the codebase's panic-on-error
pattern for DB operations), and only break the loop when err == nil && has ==
false; keep references to s.db.Has and key when making this change so the check
is: if err != nil -> panic(err); else if !has -> break.

In `@node/derivation/derivation.go`:
- Around line 703-739: The anonymous closure that calls d.l2Client.NewL2BlockV2
(producing err) is not checked before proceeding to d.l2Client.HeaderByNumber;
capture and check the returned err from the closure (the variable named err
after the closure invocation) and if non-nil return it (or wrap it with context)
instead of continuing, so failures from NewL2BlockV2 are propagated and
HeaderByNumber is only called when execData was successfully written.

In `@node/derivation/verify_local.go`:
- Around line 203-216: The docs/comments are misleading: rebuildBlob (function
rebuildBlob / call site in verify_local.go) only validates blob count and
returns rebuilt blobs; per-index hash equality is performed later in
derivation.go (the loop comparing rebuilt[i] to batchInfo.blobHashes[i]). Update
the comment above rebuildBlob to state it only checks length and returns rebuilt
blobs (not per-hash equality), and update the ErrBatchVerifyDivergence /
versioned_hash_mismatch comment mapping in verify.go to indicate divergence is
detected at the derivation.go comparison site rather than inside rebuildBlob;
adjust any comment text referencing "rebuildBlob wraps
divergence/versioned_hash_mismatch" to reflect the current behavior.

In `@node/sync/syncer.go`:
- Around line 92-113: The code calls s.fetchL1Messages() twice: once
synchronously before the goroutine and again immediately inside the goroutine
loop; remove the redundant in-goroutine immediate call by changing the goroutine
loop to wait on the ticker (or select on ctx.Done and t.C) before calling
s.fetchL1Messages() so that the initial synchronous fetch (the one before the
goroutine) is the first fetch, leaving the goroutine to poll only on t.C; update
the comment near the ticker (t := time.NewTicker / "don't wait for ticker during
startup") to reflect this behavior and ensure the goroutine still closes s.stop
on s.ctx.Done.

---

Outside diff comments:
In `@ops/devnet-morph/devnet/setup_nodes.py`:
- Around line 61-67: persistent_peers_value is missing the entry for
sentry-node-1 so the testnet won't discover it; after generating node keys, read
the node ID from node5/config/node_key.json and append an entry
"NODE_ID@sentry-node-1:26656" to the persistent_peers_value string (match the
existing comma-separated format), ensuring the new peer uses the actual node ID
you found; update the persistent_peers_value definition so it includes
sentry-node-1 alongside the existing sentry-node-0 entry.

In `@ops/docker/docker-compose-4nodes.yml`:
- Around line 484-496: The docker-compose env block for the tx-submitter service
contains a duplicate variable TX_SUBMITTER_SEAL_BATCH (present in the
tx-submitter-0 env list); remove one of the occurrences so
TX_SUBMITTER_SEAL_BATCH=true is defined only once, keeping the single canonical
entry and leaving all other TX_SUBMITTER_* variables unchanged.

---

Nitpick comments:
In `@node/derivation/derivation.go`:
- Around line 282-299: The TODO "reorg" comment indicates missing reorganization
handling in the loop comparing rebuilt vs batchInfo.blobHashes; either implement
explicit reorg handling or remove the TODO and document that the current
self-heal path is the intended behavior. If implementing reorg handling, add
code in the loop when rebuilt[i] != batchInfo.blobHashes[i] to invoke the reorg
resolution routine (e.g., call an existing or new method like handleReorg or
reconcileChain) before falling back to fetching full batch via
fetchRollupDataByTxHash and re-deriving with deriveForce(batchInfoFull);
otherwise delete the TODO and add a brief comment stating that mismatches are
handled by fetchRollupDataByTxHash + deriveForce to avoid misleading future
readers.

In `@node/derivation/static_scan_test.go`:
- Around line 18-54: walkNodeRepoSourceFiles currently uses nodeRoot :=
filepath.Dir(wd) which assumes the test is run from .../morph/node/derivation;
change it to reliably locate the repo/module root by climbing parent directories
from wd until you find a marker (e.g., a go.mod containing the expected module
name "morph-l2") or by invoking "go list -m -f '{{.Dir}}' <module>" and using
that directory as nodeRoot; update references in the function (wd, nodeRoot) so
the file walk uses the discovered root and fail the test with a clear error if
the marker/command lookup fails.

In `@ops/devnet-morph/devnet/setup_nodes.py`:
- Around line 108-110: The error message in the check for source_dir is
misleading: update the conditional handling around source_dir (the if not
os.path.isdir(source_dir) block that prints the error for node) to report only
the missing source directory (e.g., "Error: Missing source directory for {node}.
Exiting.") or otherwise include accurate context that destination was already
validated; adjust the process that calls print/fails so the message references
source_dir and node explicitly.