What's Changed
Features
- feat: support installing batch-gateway from OCI Helm chart by @yizhaodev in #299
- feature: send x-gateway-inference-objective header for GIE flow control by @lioraron in #303
- chore: improve deploy-demo docs and scripts by @yizhaodev in #326
- ci: add DCO sign-off check workflow by @lioraron in #327
- ci: skip DCO check for Dependabot PRs by @lioraron in #332
- test(e2e): expand batch scenarios and fix comment accuracy by @j-mok-dev in #336
Bug fixes
- fix: omit expires_at when output has no expiration by @j-mok-dev in #312
- fix(apiserver): align dev config tenant header with chart default by @j-mok-dev in #315
- fix(processor): reject nil task in Poller.enqueueOne by @j-mok-dev in #314
- fix(demo): wait for TokenRateLimitPolicy in LLM_NAMESPACE (deploy-maas) by @j-mok-dev in #319
- fix: restore apiserver request logging broken since logr migration by @yizhaodev in #324
- fix: allow cancel retry when batch is already cancelling by @j-mok-dev in #330
- fix: use PR author instead of actor for DCO bot skip by @lioraron in #334
Documentation
- docs: add processor TLS guide for HTTPS inference backends by @j-mok-dev in #329
- chore: update doc for release v0.1.0 by @yizhaodev in #337
- docs: clarify that SAR model name is independent of InferencePool name by @yizhaodev in #310
- docs(demo): note htpass-secret delete when changing MaaS test users by @j-mok-dev in #313
- docs: clarify batch-route vs llm-route security boundary by @j-mok-dev in #317
Full Changelog: v0.1.0-RC1...v0.1.0