Platform9 4.5 Release Notes
Below are new features, upgrades and bug fixes for the 4.5 release of Platform9 Managed Kubernetes and Managed OpenStack.
What’s Changed: Platform9 Managed Kubernetes
A notification will now appear when a cluster can be upgraded.
We added a Rule Details popup. On the Rules Dashboard under Monitoring click on any Rule to view its configuration. We added the ability to view active alarm details. When viewing the Monitoring Dashboard click on any alarm to view its details. We have added a new dashboard to Monitoring that enables you to view the built in Prometheus Rules.
Platform9 can now detect and display Cluster CSI details, CSI Driver details can now be viewed on the Cluster Details dashboard. We added an API for querying cluster CSI driver details.
Bug Fixes and Improvements
- Resolved CVE-2020-8558: Kubernetes: Node setting allows for neighboring hosts to bypass localhost.
- etcd has been upgraded to version 3.3.22.
- Added a link to Platform9 documentation for attaching BareOS nodes on the Attach a Node dashboard.
- Fixed a bug impacting pf9ctl version command.
- Fixed a bug that prevented etcd backup intervals from updating correctly.
- Fixed a bug that would inadvertently cause clusters to restart.
- Fixed an issue causing Kubelet connections to fail which resulted in nodes entering NotReady status.
- Fixed an issue impacting storage volumes for clusters running Ubuntu 18.04 on AWS.
- Fixed an issue impacting the cluster status changes during cluster creation.
- Fixed an issue that was causing incorrect Kubelet log rotation.
- Fixed an issue that was preventing AWS nodes from being removed form Platform9 on node termination in AWS.
- Fixed an issue that was preventing Kubelet from recreating the Hyperkube container after the pod was OOM-Killed.
- Fixed a bug that allowed users to delete the Service tenant.
- Fixed a bug that would allow incorrectly formatted SSH Keys to be used when creating a cluster on Azure.
- Fixed a bug that would allow Multi-Master Clusters without a Virtual IP.
- Fixed an issue impacting the deployment of the example Pod from the Platform9 SaaS Management Plane.
For the latest Kubernetes support matrix please visit Kubernetes Support Matrix Latest
What’s Changed: Platform9 Managed OpenStack
Added Python 3.6 support for all OpenStack services. With Python3.6 support, minimum version required of the OVS package on hosts is 2.11.1, which can be downloaded from the Platform9 repository.
For CentOS/RHEL hosts:
yum -y install https://s3-us-west-1.amazonaws.com/platform9-neutron/noarch/platform9-neutron-repo.noarch.rpm yum update openvswitch
For Ubuntu hosts:
wget -q -O - https://platform9-neutron.s3-us-west-1.amazonaws.com/ubuntu_latest/key.gpg | sudo apt-key add - add-apt-repository 'deb http://platform9-neutron.s3-website-us-west-1.amazonaws.com/ubuntu_latest /' apt-get update apt-get -y install openvswitch-switch
Bug Fixes and Improvements
- Added pre-requisites checking to VM-HA api so as to disallow enabling HA on host aggregates with less than 4 hosts or when any host is offline at the time of enabling HA itself.
- Fixed an issue where Cinder was not filtering Glance endpoints by region name at the time of volume creation leading to timeouts.
- Fixed an issue where keystone client used by nova, cinder and ceilometer services on hosts were not using the proxy, despite comms service being configured to reach the control plane via a proxy server.
- Fixed an issue where VMWare discovery service would fail to parse special characters in VC password.
- Fixed an issue with Aodh due to missing keystone domain id in service configuration when notifying gnocchi-based alarms’ action.
- Fixed an issue with soft-deleting VMs booted from volumes where volume would continue to be in attached state even if VM is deleted.
- Fixed an issue with stale allocations for a resource provider in nova placement database that would prevent deletion of that RP and corresponding compute service during hypervisor role de-authorization.
- Fixed an issue with VM-HA where rabbitmq queue keeps growing for a host when it is removed from the HA cluster.
- Fixed a bug that allowed users to migrate VMs to hosts outside of their allocated availability zone.
- Added a warning message to inform users that at least three hosts are required to configure OpenStack High Availability
Starting with this Platform9 release, the deprecated ceilometer-api is disabled in favor of using equivalent APIs provided by Gnocchi, Aodh and Panko services for metrics, alarms and events respectively. With this change, any references to native ceilometer metrics should be dropped from Aodh alarms and Heat stack templates. You should transition to using Gnocchi-based alarms for autoscaling feature.
Updated Ubuntu Repository
Updated existing ubuntu_latest repo with signing of files using gpg key. Download key from s3 bucket using following
wget -q -O - https://platform9-neutron.s3-us-west-1.amazonaws.com/ubuntu_latest/key.gpg | sudo apt-key add - Debian path in /etc/apt/sources.list.d/platform9*.list should be modified to following add-apt-repository 'deb http://platform9-neutron.s3-website-us-west-1.amazonaws.com/ubuntu_latest /' apt-get update
#### VMware Appliance
Added below annotation in VMware appliance template OVF suggesting it is used as Platform9 Gateway to avoid accidental deletion.
For the latest OpenStack support matrix please visit OpenStack Support Matrix Latest
September 16, 2020
Thank you for your feedback! What did you like about this article?
Thank you for your feedback! How could this article be improved?
Thank you for your feedback!