Skip to content

Instantly share code, notes, and snippets.

@derekwaynecarr
Last active January 3, 2018 19:42
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save derekwaynecarr/76b81d85d07a3840cceaed64ebe80d52 to your computer and use it in GitHub Desktop.
Save derekwaynecarr/76b81d85d07a3840cceaed64ebe80d52 to your computer and use it in GitHub Desktop.
Apps
----------------
Workloads API GA
- apps/v1 Workloads API (DaemonSet, Deployment, ReplicaSet and StatefulSet)
- batch Workloads API has a separate path to GA
API Machinery
----------------
Admission Control
* admission webhooks are now in beta
** mutation support for admission webhooks
** metrics for monitoring webhook latency
** ability to accomodate in and out of cluster webhooks
Custom Resources
* validation of custom resources (beta)
* improved documentation and samples (https://github.com/kubernetes/sample-controller)
Chunking
* Ability to fetch via API large numbers of resources in chunks to retrieve consnstent snapshot range read
* kubectl get --chunk-size={SIZE} control number of results in large lists of resources (500 batches), avoid delays and perceived latency
* ?limit=500&continue=<token>
Quota
* object count quota for all standard api objects
-- object count quota for all standard api objects
-- quota for hugepages
Auth
---------------------
Audit
* AuditEvent improved timestamps RequestRecieved and StageTimestamp
RBAC
* Aggregated cluster roles unions rules of matching cluster roles by label
AWS
---------------------
Nodes can now use instance types that use NVMe (C5)
Nodes are tainted if volumes are stuck attaching
CLI
--------------------
* kubectl get pods --field-selector=spec.nodeName=node1
Networking
--------------------
* alpha supprot for ipv6
* IPVS mode for kube-proxy to beta
Node
--------------------
* metric support for nvidia gpus, pod level local ephemeral storage
* reliability improvements to device plugins api over full lifecycle
* cpu pinning works across kubelet restarts
* cri-o is stable (passes all 1.9 e2e tests), minikube integration
* cri-containerd is beta
* various performance and reliability improvements
Scheduling
--------------------
* improvements in pod priority and pre-emption
** handles pod disruption budget
** integrated with kubelet eviction logic
* new priority function to prefer nodes that can satisfy limits in case of tie (alpha)
Storage
--------------------
Container Storage Interface (CSI) [Alpha]
- enable new volume plugins via pod deployments instead of directly mutating core kubernetes
- effort across storage orchestrators (kubernetes, mesos, cloud foundry, docker)
- enable volume plugins to become containerized in future
Raw block device support (Alpha) via Fibre Channel
Volume resizing for GCE PD, Ceph RBD, AWS EBS, OpenStack Cinder (Alpha)
Windows
--------------------
- control plane runs on linux
- kubelet and kube-proxy can run on windows server 2016 or later
- able to evaluate it for usage
- shared network namespace, reduced network complexity with single endpoint per pod
- kernel based load-balancing using Virtual Filtering Platform (VFP), analagous to Linux iptables
- CRI pod and node level statistics
Multicluster
--------------------
- federation moved out of tree
- federation specific behavior not in kubectl
- kubefed is no longer part of kubernetes
- federation servers are no longer in hyperkube
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment