- Added support for Alertmanager API v2, InitContainers in Kubernetes service discovery, and optional Snappy compression of WAL records.
- Implemented multiple performance and resource improvements: reduced WAL disk usage, lower allocation overhead in PromQL, faster query hashing, remote‑write allocation tweaks, and creation of a new clean WAL segment on startup.
- Fixed critical bugs such as a maxGauge race condition, ZooKeeper connection leak, unsafe snapshot handling, and various compaction‑related errors.