Zero 1.6

Postgres 17 Replication Failover and Performance

Installation

Copynpm install @rocicorp/zero@1.6

Upgrading

PlanetScale Replication Failover

Previous Zero versions lost replication slots after a PlanetScale Postgres failover, forcing resync. Zero 1.6 fixes this problem. To enable support, run zero-cache with ZERO_UPSTREAM_PG_REPLICATION_SLOT_FAILOVER=true. You must then do a resync, and register the full range of slots in PlanetScale's failover configuration. See High Availability and Failover for the full setup.

Features

• Litestream Region: Added support for deployments in non-standard AWS partitions like GovCloud (thanks @ericykim!).

Performance

• ~5% CPU reduction in replication-manager benchmarks by reusing stringified payloads across subscribers
• Faster EXISTS subqueries via the new Cap operator — can turn exists-heavy queries that previously timed out into ones that complete in milliseconds
• Bulk-insertion optimization in Replicache via putMany, speeding up large sync patches (3-5x faster for typical sync batches, up to 50x+ for large preloads)
• Batch deletes/upserts in SQLiteStore writes (~8x faster on 1k-put commits)
• Batch concurrent SQLiteStore get/has reads into fewer database queries
• Parallelize I/O during pull and rebase
• Heap-based k-way merge in fetchMergeSort (O(log K) per row vs O(K))
• Initial sync progress reporting uses pg_class estimates instead of full scans
• De-dupe SQLite requests in flip-join when children want the same parent
• zero-sqlite3 now gathers up to 128 STAT4 samples per index, improving SQLite query planning for skewed indexed data

Fixes

• Returning to an app after stale-tab GC or CVR purge caused a full page reload
• "Row already exists" assertion failures during poke processing
• Deadlock during post-initial-sync changeLog reset
• Zombie ViewSyncers inflated active-client-groups metric
• ConcurrentModificationException now reconnects instead of erroring
• zero-cache startup errors during change-streamer init not published
• TypeError: Expected string at context.query. Got null
• Replication slots were lost after a PlanetScale Postgres failover
• Replication slot creation timeouts crashed the server during backfill retries
• Shadow sync threw when a synced table could not be queried by ZQL
• Shadow sync now reports metrics and logs verification-success counts
• WebSocket errors are now logged as warnings instead of errors
• Inspector now reports last query hydration time instead of histogram
• Query/mutator functions now allow omitting args
• Reoduce log volume at INFO level
• Reclassify common Postgres config errors as warnings instead of errors
• Union-fan-in queries could ignore reverse ordering
• Abort zero-cache when ChangeDB CDC tables go missing
• CVR purge failures retried immediately in a tight loop instead of backing off

Breaking Changes

None.