Last checked and updated on 10 Jun 2026
What Is New in Kubernetes 0.8
Kubernetes 0.8 delivers a major infrastructure overhaul and introduces powerful new orchestration primitives. This release focuses on making the system more robust, observable, and production-ready.
| Category | Key Changes |
|---|---|
| New Features | Pet Set (alpha), Cluster DNS Addon, Node Controller |
| API & Client | API Groups, Go Client Library, Kubectl Improvements |
| Networking | Service Proxy Changes, kube2sky Integration |
| Storage | Persistent Volume Recycling, NFS Volume Support |
| Core & Reliability | High-Availability Master, Pod Lifecycle Redesign |
How did Kubernetes 0.8 improve stateful application management?
The headline feature is the alpha introduction of Pet Set, the precursor to StatefulSet. This finally provides a native way to manage stateful workloads like databases with stable network identities and persistent storage.
Pets have stable names (pet-0, pet-1) and dedicated storage that follows them around, even if they get rescheduled to a different node. This is a game-changer for running anything that isn't stateless. We also got Persistent Volume recycling, which automatically cleans up volumes for reuse, making dynamic provisioning more practical.
What networking and service discovery enhancements were made?
Service discovery became a first-class citizen with the new cluster DNS addon. It runs a SkyDNS server that automatically creates DNS records for every Kubernetes service, letting pods find each other by name instead of relying on environment variables.
Under the hood, the service proxy was completely rewritten. The userspace proxy was replaced with a more efficient iptables-based implementation, which drastically reduces latency and resource overhead. This makes service networking faster and more reliable at scale.
How did the API and client experience evolve?
This release laid the groundwork for API longevity with the experimental API Groups feature. It organizes the API into groups and versions, preventing future changes from breaking existing clients-a crucial step for a stable platform.
For developers, the official Go client library (k8s.io/client-go) became the recommended way to
interact with the API programmatically. The kubectl CLI also got smarter with better output
formatting and the new rolling-update command for managing deployments.
What core architectural changes improved reliability?
High-availability for the master components is now achievable. You can run multiple replicas of the API server behind a load balancer, eliminating the single point of failure that plagued earlier setups. This is essential for any serious production cluster.
The entire pod lifecycle was redesigned to be more deterministic. The Kubelet's internal state management was rewritten to be more resilient to crashes and restarts, ensuring pod states don't get out of sync with reality.
FAQ
Is Pet Set production-ready in 0.8?
No, it's explicitly an alpha feature. The API and
behavior are unstable and likely to change. Use it for experimentation but not for critical workloads yet.
Do I have to use the new Cluster DNS addon?
No, it's optional but highly recommended. The
older environment variable-based service discovery method still works, but DNS is more flexible and
Kubernetes-native.
What happens to my existing services with the new iptables proxy?
They should work
seamlessly.
The change from the userspace proxy is transparent to your applications. You'll just see lower latency and
better
performance.
How do API Groups affect my existing API calls?
For now, the legacy API path
(/api/v1) remains fully supported and unchanged. API Groups introduce new paths
(/apis/groupName/v1) for future extensions, so your existing code won't break.
Can I set up a multi-master HA cluster easily?
The components are there, but the setup is
still largely manual. You need to configure multiple API servers, etcd replicas, and a load balancer yourself.
Automated tooling for HA came in later releases.