Two latent apply-path hazards from the ScadaBridge adversarial architecture
review. Neither has a live trigger in any current consumer — ScadaBridge's ten
replicated tables carry no foreign keys at all (verified against the running rig's
site-localdb.db) and OtOpcUa's LocalDb tables carry none either — but both are the
kind that surface as silent divergence rather than as an error, so they are closed
before a consumer's schema grows into them.
Wire-compatible: no proto change, no lib_schema_version bump, no on-disk schema
change. No consumer code change.
- RegisterReplicated now refuses a table that is the PARENT of a foreign key with
a row-mutating ON DELETE action (CASCADE / SET NULL / SET DEFAULT), naming the
offending child column and action. The apply path writes remote rows with
INSERT OR REPLACE, whose PK-conflict resolution DELETES the existing row first;
connections run PRAGMA foreign_keys=ON, so that implicit delete fires the child's
ON DELETE action while the capture guard is raised. The child rows would change
on the applying node only, produce no oplog entry, and diverge the pair with
nothing able to detect it — LWW compares only rows it was told about. RESTRICT /
NO ACTION stay allowed: they abort the statement instead of mutating rows behind
capture's back, which the apply path already handles as per-entry poison. The
check inspects every foreign key in the database and runs at registration;
SQLite cannot add a foreign key to an existing table, so a child table created
later is caught at the next startup's registration, not at the DDL. Documented
on ILocalDb.RegisterReplicated and in the README's limitations list.
- An entry's row write and its ledger write are now atomic with each other. They
are two statements and SQLite aborts only the failing one, so a statement-level
fault between them — the case the per-entry poison catch exists for — committed
a half-applied entry: a tombstone whose table row was gone while
__localdb_row_version still called it live (SnapshotStreamer can only warn and
skip that, so the peer never learns the row died), or an upsert whose new row
carried the OLD version and therefore lost the next comparison it should have
won. The pair now runs inside a per-entry SAVEPOINT, rolled back (and released,
so nothing nests) before dead-lettering, leaving both writes or neither while
the rest of the batch proceeds and commits.
Every new test was run against the unfixed code first and fails there with exactly
the divergence shape described: the tombstone case reads back null instead of the
live row, the upsert case reads back the remote row against an unchanged ledger,
and the registration cases throw nothing.
Suite: 213/213 green, build 0 warnings / 0 errors.
WP3.3 of the ScadaBridge architectural-review remediation. Six changes, all
wire-compatible with 0.1.x peers — no proto change, no lib_schema_version bump. The
on-disk bookkeeping schema goes to v2 and is upgraded in place when a file is opened.
DeregisterReplicated(table). There was no way to stop capturing once a file had been
registered, so a node that does not replicate kept paying three triggers inside every
write transaction, appending to an oplog nothing would ever drain. Drops the triggers and
prunes the table's oplog + row-version rows in one transaction; idempotent, and
deliberately usable on a table this process never registered, since stale triggers left by
an EARLIER build are the primary case. Deregistration must be symmetric across a pair, and
already is enforced: the handshake compares registered-table digests fail-closed, so a node
that drops a table its peer still replicates stops syncing loudly rather than diverging.
RegisterReplicated(table, baselineExistingRows: true). Closes the late-opt-in hole: the
snapshot streamer pages from __localdb_row_version, which only triggers populate, so a node
that started replicating LATE never sent its pre-existing rows — silently, forever. Seeding
alone would not have fixed it, because a node whose oplog starts at seq 1 presents no gap at
the handshake and is told it needs no snapshot; the seed therefore also flags needs_snapshot,
and the existing resync machinery carries the rows. The floor is HLC 0 stamped with this
node's id: every genuine HLC is a UTC millisecond shifted left 16 bits and so is strictly
greater, which makes a baselined row lose every comparison against a real remote write and
win only where the peer has no version of that key at all. Two nodes baselining the same key
both hold 0 and resolve by the node-id tie-break — convergent, and nothing better is
available for two rows of unordered antiquity. Opt-in, so the documented CDC contract is
unchanged unless asked. Pruning the ledger on deregister is safe precisely because
re-registration can re-baseline; the two APIs are a pair.
O(1) backlog depth. GetOplogDepth ran COUNT(*) over the entire unacked backlog, once a
minute from maintenance AND on every metrics scrape — a full scan of a million rows exactly
when a node is already behind. Now MAX(seq) - MAX(last_acked_seq, MIN(seq) - 1), floored at
0: three b-tree seeks, measured at 58 VM steps against 100,020 for the count on a 50k
backlog. Exact, because seq is AUTOINCREMENT and every prune path removes a prefix (ack
pruning by watermark, the row cap by count, the age cap by an HLC cutoff that rises with
seq). The MIN clamp is what keeps it exact after a cap prune has moved the horizon above a
stalled watermark — without it the caps would re-flag on every tick forever.
Byte-budget batching. MaxBatchSize was the only cap on an outbound message, which made it a
proxy for a byte budget that consumers had to tune against their widest row: ScadaBridge
pinned it to 16 because 70 KB config_json times the 500 default is ~35 MB against a 4 MB
gRPC limit. MaxBatchBytes (default 2 MB) now bounds delta and snapshot messages by summed
serialized size, with the row count demoted to a secondary cap. A row larger than the budget
is sent alone rather than stalling the stream on it. Since more, smaller messages would
otherwise mean more apply transactions and more acks, inbound delta batches and acks ALREADY
queued are coalesced — one transaction and one ack per burst, bounded by the sender's own
caps, and a no-op under a trickle.
LwwApplier prepares each statement once. It went through the generic transaction façade,
which builds a fresh SqliteCommand per call: three SQL parses per applied row, on the path
that drains a reconnect backlog. Statements are now held for the batch and rebound.
Dropped __localdb_oplog_hlc. It indexed (table_name, pk_json, hlc), which no query has ever
used — every oplog access path is by seq or a full hlc-predicate delete — while costing a
b-tree insert inside every captured consumer write.
184/184 tests pass (was 149). New coverage: schema upgrade drops the index and preserves
identity/watermarks/data, deregistration idempotency and trigger absence and cross-process
cleanup, baseline seeding at the floor with LWW safety proved through the real applier,
depth asserted equal to the counting query across every prune scenario plus a constant-time
assertion over a 200k backlog, byte-budget chunking including the one-giant-row edge on both
the delta and snapshot paths, prepare-once statement counting, and coalescing that must not
swallow the message that ends its run.
Version 0.2.0.
Joseph Doherty