Base Command Manager / Bright Cluster Manager Release Notes

Release notes for Bright 8.1-7

== General ==

- New Features

* Expose Kubernetes dashboard without the need of a local proxy

- Fixed Issues

* pdsh -v crash when cmdaemon is down

== cmdaemon ==

- New Features

* Configurable behavior of prejob healthchecks

- Improvements

* Reduce the repeated logging that prometheus is disabled
* Allow a boot role to serve only a subset of images
* Speed up device manager hostname based lookup

- Fixed Issues

* An issue with creating an alias on tagged vlan interfaces
* In some cases, null address in /etc/resolv.conf written by node-installer
* In some cases, an issue with migrating new users to openstack

== Bright View ==

- New Features

* List of available metrics, labels and values for PromQL

- Improvements

* Custom app configuration
* Performance improvements in monitoring

== cm-cluster-extension ==

- Fixed Issues

* CX-AWS: crash of cm-cluster-extension when selecting a vpc with no subnets

== cm-setup ==

- Fixed Issues

* Increased speed at which node selection page is loaded

== cmsh ==

- Fixed Issues

* An issue with cmsh device --role and --overlay selectors
* cmsh device option intersection with empty list does not result in an empty list

== cod ==

- New Features

* COD-OS: speed up "cluster list", and a few other subcommands, by optimizing the way the tool reuses keystone sessions (this also reduced the load on Keystone APIs due to fewer auth requests)

- Fixed Issues

* Improved handling non-ascii characters passed in as an argument
* New subcommand 'image stats' producing head node images stats for the current deployment
* COD-OS: Non backward-compatible change to naming policy of cod routers. When a COD_PREFIX is in use, the cod-router(s) need to be renamed to $COD_PREFIX-router.

== monitoring ==

- Fixed Issues

* An issue with health check data interpolation when there is 'no data'
* In some cases, job information does not correctly handle the time zones