Platform9 3.7 release notes
Platform9 Managed Kubernetes
1. Bug fixes
The following bugs have been resolved in the latest release of Platform9 Managed Kubernetes.
- Resolve issue with secure host certificates not being issued when simultaneously attempting to authorize a large number of hosts.
Platform9 Managed OpenStack
1. Neutron project upgraded to Pike release
OpenStack Neutron has been updated to the Pike release. This release brings a number of new features, critical bug fixes, and stability enhancements.
2. Telemetry services upgraded to Pike, and Panko introduced
Gnocchi, the time series database designed to store time series data generated by Ceilometer, has been updated to version 4.2.5.
OpenStack Telemetry projects (Ceilometer and Aodh) have been updated to the Pike release. This release also introduces support for the Panko project. Panko is a component of the OpenStack Telemetry project which provides scalable storage for event data generated by Ceilometer.
Users leveraging automation against Ceilometer are encouraged to migrate to the updated counterparts in aodh (alarms), Gnocchi (metrics), or Panko (events). Additionally, Heat Orchestration Templates leveraging auto-scaling should be updated to utilize the aodh alarm type.
3. Clarity allows booting Instances from volume by default
It is now possible to configure Clarity to default to booting instances from volume during Instance creation. Contact Platform9 if you would like to enable this setting.
4. Single sign-on support for SAML ECP (Enhanced Client or Proxy)
Platform9 now supports SAML authentication using the ECP (Enhanced Client or Proxy) standard, in addition to the previously supported WebSSO sign-in method. This allows non-browser clients (API/CLI) to authenticate with Platform9 using SAML authentication.
5. End-of-life for Ubuntu 14.04
Platform9 release 3.7 officially drops support for Ubuntu 14.04 LTS as a host operating system for nodes running Platform9 Managed OpenStack. Operators must deploy or upgrade to Ubuntu 16.04 LTS before being eligible to upgrade to Platform9 3.7.
6. Bug fixes
This release also contains a number of bug-fixes which should result in a better user experience for your Platform9 cloud platform! Some significant ones are listed below.
- Resolved issue related to Nova where Nova incorrectly tracks live migration progress. https://bugs.launchpad.net/nova/+bug/1727845.
- In environments utilizing Highly Available Image Libraries, Nova now gracefully handles situations where an image is not present on the initially queried Image Library host. Nova will now attempt to locate the image on alternative image library hosts, and return an error if the image cannot be located throughout the HA cluster.
- Fixes an issue wherein an image which is placed in the image discovery directory during pf9-imagelibrary startup may be deleted if the name of the discovered image matches that of a previously deleted image.
- Stability and reliability fixes to VM-HA.
7. Known issues
Deleting an Instance on VMware will not remove associated disk images if the Instance previously underwent Storage vMotion. The Nova VMware driver creates Instances, and their associated disk images, using the OpenStack UUID as the name. To assist with identifying the Instance in vCenter, Platform9 OpenStack renames the Instance to match its configured hostname. However, the filename of the disk on the datastore remains the Instance’s UUID. When an Instance is storage vMotioned, the migration will rename the Instance files on the datastore according to the name of the Instance. This breaks Nova’s association to the disk image, resulting in the inability to remove the disk when the Instance is deleted.
Floating IPs cease to function in Distributed Virtual Routing (DVR) environments for Instances on CentOS/RHEL 7.5 hypervisors running kernel version 3.10.0-862.3.2. See https://bugzilla.redhat.com/show_bug.cgi?id=1572983 for more information. This kernel bug has been fixed in version 3.10.0-898.el7. The solution is to upgrade the kernel to this or a newer version.
8. Known limitations
For a HA cluster with 5 consul
servernodes, we can tolerate up to 2
servernode failures, in addition to any
clientnode failures. See https://www.consul.io/docs/internals/consensus.html#deployment-table for details. After an HA evacuation event, if the
consul memberscommand output on a host shows only 3 server nodes alive, please reach out to Platform9 Support to disable/re-enable HA on that cluster.
VMware live migration not supported when Instance has attached Cinder volumes. The VMware Cinder driver creates ‘shadow VMs’ when creating VMDK volumes. Cinder manages the volume using this shadow VM. If an Instance with attached Cinder volumes is migrated, and the migration also results in a datastore migration, the migration will corrupt the shadow VM. The VMDK associated with shadow VM will be migrated to the new datastore, and renamed according to the name of the VM, thus corrupting original reference to the shadow VM.
August 31, 2018