1.8.4-tectonic.1

Tectonic 1.8.4-tectonic.1 (2017-12-19)

Core Components

• Updates to Kubernetes v1.8.4
• Updates to Docker 17.03
  • With this release and going forward, Tectonic will manage the version of the Docker Engine that's installed on the platform and automatically update it to the most recent validated release. We ship Docker 17.03 with Tectonic 1.8 and it's been fully tested for security and stability.

Tectonic Installer

• Enable flannel hairpin mode

Tectonic Open Cloud Services

• Read the introduction blog post
• New Vault Open Cloud Service
  • Install and manage instances of a highly available secret store
  • Ability to enable specific namespaces to run Vault
• Enhancements to the etcd Open Cloud Service
  • Visualize key etcd metrics
  • Ability to enable specific namespaces to run etcd
  • General user interface improvements
• Enhancements to the Prometheus Open Cloud Service user interface
  • Ability to enable specific namespaces to run Prometheus
  • General user interface improvements

Tectonic Console

• Ability to jump to open alerts from the main dashboard
• Ability to jump to crash looping pods from the main dashboard
• Improves the accuracy of CPU and other cluster health gauges
• Enhanced search page with filtering
• Fixes bug when editing Node labels

Tectonic Monitoring

• Updates to Prometheus 2.0. Read the overview on the CoreOS blog.
• Dramatically reduced resource usage

Known Issues

• The Service endpoint that routes to the API Server(s) may contain stale entries when setting the --apiserver-count flag greater than one. The CoreOS engineering team has merged code to fix this behavior, which will be available in Kubernetes 1.9.
  • Upstream issue: kube-apiserver endpoint cleanup when --apiserver-count>1
  • Upstream issue: add apiserver-count fix proposal
• External etcd clusters using hostnames instead of IPs may be affected by a breaking change in etcd v3.2 flags
• Pods that contain initContainers will have a 300 second grace period upon deletion, which is a new Kubernetes default.

Upgrading to 1.8.4-tectonic.1

Upgrading to 1.8.4-tectonic.1 requires first upgrading to 1.7.9-tectonic.3.

Changes between "minor" (vs "patch") versions of Kubernetes are controlled by the Tectonic update channel. After upgrading to the 1.7.9-tectonic.3 release, select the production or pre-production Tectonic-1.8 channel and then click "Update" to start the rolling, no-downtime upgrade process.

ThirdPartyResources (TPRs) were deprecated in Kubernetes 1.7 and replaced by CustomResourceDefinitions (CRDs). As of 1.8, TPRs have been completely removed from 1.8. Transition all TPRs within your cluster to CRDs before upgrading to Tectonic 1.8.x. If TPRs are present, your upgrade will pause until they are removed.