Base Command Manager / Bright Cluster Manager Release Notes
Release notes for Bright 8.2-17
- Improvements
* Update of cm-openssl to 1.1.1
* We recommend that all Bright packages are updated at the same time as cm-openssl
== cmdaemon ==
- New Features
* New node-installer SKIP install mode to prevent checking or syncing the disk
* New feature to allow all head nodes to be included as DHCP servers even if they have no interface on a network
- Improvements
* Allow regex to be specified for exclude list in /proc/mount sampler
* New option to allow changing the node IPs without rebooting
* Write active head node url in /var/run/cmd.url, so that tools running as non-root can access it
- Fixed Issues
* Display a warning about node exclude list being used without the node image being explicitly set
* PromQL over Bright metrics need to use raw and consolidated data
* Rare crash in sample now
* An issue with upgrading clusters with docker registry roles
* /etc/genders does not support two sets of [] ranges
* In some cases, named on the passive sles12sp4 head node does not run or restart
* Entity measurable cache is cleared before full stop, which in some cases can result in missing monitoring data after cmd restart
* An issue with handling LSF dynamic hosts
* An issue with Open vSwitch (OVS) interface population, hypervisors will clear OVS database on reboot, to avoid potential population errors
* In some rare cases, failure during AWS VPC creation
* Data Node is enabled for node categories after importing XML
* In some cases after cmha-setup finalize secondary cmdaemon remains DOWN
== node-installer ==
- New Features
* Do not allow diskless nodes with install nosync
- Fixed Issues
* Node installer crash when /certificates is mounted read-only
* Improved node-installer IB hardware type detection.
== cm-kubernetes-setup ==
- Fixed Issues
* An issue with managing Kubernetes users in the case of external LDAP
== cm-lite-daemon ==
- Improvements
* Switch cm-lite-daemon to python 3.6 or higher
- Fixed Issues
* cm-lite-daemon IO metrics need to be cumulative
== cmsh ==
- Improvements
* Display overlay roles in cmsh category role mode
* New --status flag to the cmsh device list command
- Fixed Issues
* An issue with cmsh provisioningnodemapping command giving the wrong overview in the case of cloud
== pythoncm ==
- Improvements
* Allow optional timeouts for pythocm monitoring calls
* Set correct netwok interface priorities for interfaces created with pythoncm
- Fixed Issues
* An issue with pythoncm grab_image when an image instance is supplied
== slurm ==
- Improvements
* Upgrade slurm18 to 18.08.9 and slurm19 to 19.05.5