Skip to content

v0.11.0-rc5

Pre-release
Pre-release

Choose a tag to compare

@arnaubennassar arnaubennassar released this 10 Aug 14:54
· 60 commits to develop since this release
e716ceb

πŸš€ What's New

  • bridgesync from_address/network_ids bridge-listing timeout fix (#1780): GET /bridge/v1/bridges was full-scanning the ~1.1M-row bridge table whenever from_address was supplied, because UPPER(from_address) = UPPER($n) had no usable index. New migration bridgesync0016 adds an expression index on UPPER(from_address) matching the query predicate exactly.
  • bridgetracker API docs, fail-fast startup, and rate-limit cleanup (#1784, closes #1781, #1782, #1783): rewrote docs/bridgetracker/API.md to match the real wire format (most enums are bare strings; only ErrorStep.error_type/CertificateData.status use numeric+_string; event nesting, omitempty optionals, and the never-populated expected_duration are now documented accurately). The proxy now fails fast at startup if [Tracker].L1GlobalExitRootAddress is unset instead of silently stalling every L1β†’L2 route at WaitingGERUpdate. MaxRequestsPerIPAndSecond now defaults to 0 (unlimited) in [PublicREST]/[AdminREST]/proxy [REST], where it was never enforced anyway; [RPC] keeps its default of 10 since that section is enforced via tollbooth in github.com/0xPolygon/cdk-rpc.
  • SQLite query-planner regression fix: dropped idx_bridge_destination_network (added, unreleased, by #1780's migration) after it was found to hijack the planner on a 1M-row fixture β€” aggkit never runs ANALYZE, so the low-cardinality index stole the plan from idx_bridge_deposit_count_desc and idx_bridge_from_address_upper, regressing network-filtered paging ~0.5ms β†’ ~600ms. Folded into bridgesync0016 before its first release, so no extra migration is needed.

⚠️ Breaking Changes

  • πŸ› οΈ Config: running the proxy with the tracker component now fails at startup if [Tracker].L1GlobalExitRootAddress is unset/zero (previously it started and silently never resolved any L1β†’L2 bridge). Deployments running the tracker must set the address (#1784).
  • πŸ› οΈ Config: MaxRequestsPerIPAndSecond default changes from 10 to 0 (unlimited) in [PublicREST], [AdminREST], and proxy [REST]. Behavior is unchanged in practice β€” these sections never enforced the limit β€” but explicit configs pinning 10 there should be reviewed. [RPC]'s default stays 10 and remains enforced (#1784).
  • πŸ—„οΈ Storage: new migration bridgesync0016 (expression index idx_bridge_from_address_upper on UPPER(from_address)) applies automatically on startup (#1780).

πŸ“‹ Config Updates

  • 🧾 Tracker (proxy) β€” now required when running the tracker component:
    [Tracker]
    L1GlobalExitRootAddress = "0x..."   # <- now required; startup fails fast if unset
  • 🧾 REST rate limiting β€” new default, unlimited where the field was never enforced:
    [PublicREST]
    MaxRequestsPerIPAndSecond = 0   # was 10; unlimited (never enforced here)
    
    [AdminREST]
    MaxRequestsPerIPAndSecond = 0   # was 10; unlimited (never enforced here)
    
    [RPC]
    MaxRequestsPerIPAndSecond = 10  # unchanged; IS enforced (tollbooth via cdk-rpc)

πŸ› Bug Fixes

  • GET /bridge/v1/bridges no longer times out when filtering by from_address on large bridge tables (#1780).
  • docs/bridgetracker/API.md now matches the actual wire format (#1781/#1784).
  • Proxy tracker component fails fast on missing L1GlobalExitRootAddress instead of silently stalling (#1782/#1784).
  • Removed a low-cardinality SQLite index (idx_bridge_destination_network) that regressed bridge-paging query performance (#1784).

Full changelog: v0.11.0-rc4...v0.11.0-rc5