LeaderWorkerSet Patch release 0.4.1

This is a patch release for v0.4.0

Leaderworkerset v0.4.0

Features:

• Support specify NetworkConfig
• Support leader elect for lws controller
• Add group size as an environment variable
• Add an example for using llama.cpp to deploy a distributed inference service
• Update the GPU multi-node inference with vLLM example to serve Llama3.1-405b model

What's Changed

• Update the lws version of the installation guide by @liurupeng in #160
• Bump the kubernetes group with 6 updates by @dependabot in #165
• add leader elect args by @gujingit in #163
• Add executable permission to the example script ray_init.sh by @xgchena in #166
• docs: add LeaderWorkerSet API reference by @zmvictor in #171
• Bump the kubernetes group with 6 updates by @dependabot in #175
• Add example for llama.cpp by @justinsb in #174
• statefulset finalizer rbac update by @kannon92 in #176
• add verify job and fix code where it fails by @kannon92 in #180
• Bump github.com/onsi/gomega from 1.33.1 to 1.34.0 by @dependabot in #181
• set golangci-lint for 15 minutes to be same as jobset by @kannon92 in #183
• Bump github.com/onsi/ginkgo/v2 from 2.19.0 to 2.19.1 by @dependabot in #182
• Fix missing startupPolicy in applyConfiguration by @kerthcet in #186
• Bump github.com/onsi/gomega from 1.34.0 to 1.34.1 by @dependabot in #187
• updating vllm docs to use llama3 405B as example by @Edwinhr716 in #185
• No worker sts when size=1 by @kerthcet in #190
• Bump cert-controller to v0.11.0 by @kerthcet in #191
• Bump github.com/onsi/ginkgo/v2 from 2.19.1 to 2.20.0 by @dependabot in #192
• add golangci config to project by @kannon92 in #194
• Headless Service Per Replica KEP by @Edwinhr716 in #188
• fix make generate by @kerthcet in #196
• Bump the kubernetes group with 6 updates by @dependabot in #198
• We should not import from k8s/kubernetes. by @kannon92 in #200
• Updated vLLM yaml example to use LWS_LEADER_ADDRESS by @Edwinhr716 in #202
• Bump github.com/onsi/ginkgo/v2 from 2.20.0 to 2.20.1 by @dependabot in #201
• Bump github.com/onsi/gomega from 1.34.1 to 1.34.2 by @dependabot in #205
• Bump github.com/onsi/ginkgo/v2 from 2.20.1 to 2.20.2 by @dependabot in #206
• Bump kubernetes dependencies to v0.31.x. by @mbobrovskyi in #207
• feature: Add group size environment variable injection by @googs1025 in #208
• simply llama.cpp example by @carlory in #212
• Make llama.cpp example become avaliable on macOS by @carlory in #211
• Adding SubdomainPolicy to support a service per replica by @Edwinhr716 in #197
• Deprecated DefaultRestartPolicy with NoneRestartPolicy by @kerthcet in #214

New Contributors

• @xgchena made their first contribution in #166
• @zmvictor made their first contribution in #171
• @justinsb made their first contribution in #174
• @mbobrovskyi made their first contribution in #207
• @carlory made their first contribution in #212

Full Changelog: v0.3.0...v0.4.0

Leaderworkerset v0.3.0

Features:

• RollingUpdate with MaxSurge support
• Subgroup support for disaggregated serving
• Example for multi-node serving of llama 70B on GPUs with vLLM
• Add a new start policy API
• Inject leader address environment variable to every container
• Spec.rolloutStrategy should be a non-required field

Acknowledgments

Thanks to our contributors in this release, in alphabetic order:
@ahg-g @Edwinhr716 @googs1025 @gujingit @jjk-g @kerthcet @liurupeng @nayihz

Leaderworkerset v0.2.0

Features:

• Support RollingUpdate with MaxUnavailable
• Allow Prometheus to gather metrics gathered by controller-runtime
• Fix TPU env var assignment when leader pod doesn't request TPU
• User guide to deploy multi-host inference with Saxml
• Increase qps limit for pod scheduling
• Setup E2E test and improve test coverage

Acknowledgments

Thanks to our contributors in this release, in alphabetic order:
@ahg-g @Bslabe123 @Edwinhr716 @googs1025 @kannon92 @kerthcet @liurupeng @nayihz @Zeel-Patel

Leaderworkerset v0.1.0

Features:

• Support creating groups of pods as a unit
• Support dual-template, one for leader and one for the workers
• Support autoscaling through HPA
• Support topology-aware placement
• Support all-or-nothing restart for failure handling

Acknowledgments

Thanks to our contributors in this release, in no particular order:
@liurupeng @Edwinhr716 @kerthcet @ahg-g