v17.3.0rc1

Stability improvements

• Retry DNS resolution faster on startup. Speeds up connection establishment in certain scenarios (particularly Kubernetes automatic DNS provision).
• In memory mode wants SOCI transactions. Fixes a rare atomicity bug that can prevent successful crash-recovery of a captive core.
• Herder tracking cleanup and tests
• Additional startup checks
• SCP / Quorum utils: Minor fixes to reduce dependencies
• Add unit tests for application setup
• Minor logging changes:
  • Remove herder bogus recovery timer
  • Fix the logging logic of LoadGenerator to take care of the division-by-0
  • Fix bucket counters.

New features

• Add encode-asset command for debugging
• Groundwork for CAP-0038, not yet used:
  • Add Change trust XDR for pool share trustlines and related changes.
  • Add ClaimAtom
  • Big square root
  • Add LiquidityPoolEntry SQL and related changes

Breaking changes

• Store LedgerEntry in trustline table. Marked as a breaking change for visibility: no code or configuration needs to change, but on upgrade stellar-core will perform an automatic SQL schema migration, and will be unavailable to sync with its peers until this migration completes. On a fast SSD running on SQLite, the migration may be as quick as 2 minutes; on a slow spinning disk running on postgres it may take a half hour or more. Nodes should therefore be upgraded gradually and individually to avoid disruption.