# OpenFeature core/v0.13.3

2026-02-09 · [Original release notes](https://github.com/open-feature/flagd/releases/tag/core/v0.13.3) · [Full summary](https://ratatosk.io/en/releases/openfeature/core/v0.13.3)

Breaking changes 1 · Important fixes 1

## Breaking changes (1)

### Store `Watch` error logging changed

- Affects: the store's `Watch` function (Inferred)

Previously, flag-selection and watch-context errors were all logged at Error level with the generic message `error watching flags: %v`; now selection errors use `error getting flags`, while cancellation or deadline errors are logged at Debug level and other watch-context errors at Error level, with selector and error recorded as fields. The watcher is still closed when these errors occur.

Evidence: “if errors.Is(err, context.DeadlineExceeded) || errors.Is(err, context.Canceled) { + s.logger.WithFields(zap.String("selector", selector.ToLogString()), zap.Error(err)).Debug("context cancellation while watching flags") + } else { + s.logger.WithFields(zap.String("selector", selector.ToLogString()), zap.Error(err)).Error("context error watching flags")” https://github.com/open-feature/flagd/pull/1865

> * Enhance error logs in store's Watch func ([#1865](https://github.com/open-feature/flagd/issues/1865)) ([a58a707](https://github.com/open-feature/flagd/commit/a58a7076ac4aef66a10dee7a40aa2ee4b53c7169))

## Important fixes (1)

### HTTP request duration and response size histogram buckets corrected

- Affects: setups using the HTTP request duration or HTTP response size histograms from the telemetry metrics recorder (Inferred)

The request duration histogram now uses buckets [0.005, 0.01, 0.025, 0.05, 0.1, 0.25, 0.5, 1, 2.5, 5, 10] seconds, and the response size histogram uses `prometheus.ExponentialBuckets(100, 10, 8)`.

> * correct parameter order in histogram bucket configuration :warning:  ([#1859](https://github.com/open-feature/flagd/issues/1859)) ([335af32](https://github.com/open-feature/flagd/commit/335af32b6f1087d624b77ffb7b50dea612ef234f))

---

Summaries are produced by AI, so they can be wrong, and they are provided without warranty.
