Base Command Manager / Bright Cluster Manager Release Notes

Release notes for Bright 8.1-8

== General ==

- New Features

* Fixed an issue in Spark toree <> Jupyterhub integration

- Improvements

* Update of core OpenStack Pike packages
* An upgrade of already installed OpenStack Pike packages will require packages from rdo_qemu_ev repository
* The repository definition, which is disabled by default and should be enabled for the upgrade, can be found in the software image, or through contacting Bright support

- Fixed Issues

* Support deploying Kubernetes on both HA headnodes
* sview missing in slurm deb packages

== cmdaemon ==

- New Features

* Add raw value to string conversion routes for pythoncm

- Improvements

* OpenStack: HAProxy on the head and controller will now be configured with a much higher (1800s) timeout for client/server connections for novnc proxy connections. This makes idle novnc sessions in horizon not be dropped so easily.

- Fixed Issues

* Possible cmdaemon crash while creating certificates if the filesystem has wrong permissions
* New command line flag to exernal-user-cert script to allow overwriting of existing certificates
* In some cases, cmdaemon crash when accessing a file with the read file API
* An issue with checking all cert.pem paths when ExtraUserCertificatePath is specified
* Provisioning sometimes hangs on /tftpboot in HA setup
* Fixed mutex in session event handling that can lead to crashes
* Incorrect UGE and OGS drain status
* An issue with the displayed number of running cloud jobs in cmsub description
* An issue with running cm-user-init python script from within cmdaemon

== cm-cluster-extension ==

- Fixed Issues

* An issue with cm-cluster-extension to azure not working on clusters with uppercase letters in their hostname

== cm-create-image ==

- Fixed Issues

* cm-create-image failure when using -s -m options

== cm-openstack-setup ==

- Improvements

* Increased the requirement for minimum amount of memory for controller nodes from 4GiB to 8GiB

- Fixed Issues

* In some cases, an issue with OpenStack deployment failing to add sec group rules, which affects the ability to start vnodes

== monitoring ==

- Improvements

* Extend monitoring trigger resource syntax to support or/and

- Fixed Issues

* An issue with the interpolation of accounting & reporting queries when some data is missing