Base Command Manager / Bright Cluster Manager Release Notes

Release notes for Bright 8.2-5

== General ==

- New Features

* Kubernetes support for Edge clusters

- Improvements

* Upgrade keras to version 2.2.3

- Fixed Issues

* Harbor can now be used by docker when both are on the head node

== cmdaemon ==

- Improvements

* Changed the default OpenStack password_hash_rounds from 12 to 4. Only affects new deployments and newly created users (to take effect, existing users will need to re-set their passwords)

- Fixed Issues

* slurmd was not restarted after gres are updated
* Fixed rare crash in the cmdaemon, which in some cases can cause OpenStack deployment failures
* Ensure passive head node is not at the beginning of LSF_MASTER_LIST
* An issue with 1-controller-node OpenStack deployments sometimes failing when committing a Keystone object due to CURLcode 52 or 55
* cmsh hangs caused by failing to power on OpenStack cloud nodes

== cm-beegfs-setup ==

- Fixed Issues

* In some cases, an issue with selecting the correct software image

== cm-kubernetes-setup ==

- Improvements

* Added skip parameters for reboot, image-update, and package installs

- Fixed Issues

* Improved disk setup handling, which allows docker thin pool devices to also be partitions

== cmsh ==

- Fixed Issues

* In some cases, cmsh can be disconnected unexpectedly

== cmsub ==

- Fixed Issues

* An issue with setting expiration date for S3 objects

== cod ==

- Improvements

* Renamed the doubly-negated "--no-no-access-validation" flag to "--access-validation". Singly-negated flag "--no-access-validation" remains unchanged.
* New advanced option "--weak-password-policy" superseding "--weak-password-warning", allowing for fine-tuning the behavior of COD when SSH password access is enabled and a weak password is provided during cluster creation.

== wlm-setup ==

- Fixed Issues

* An issue with PBS Pro setup when offloading the server to a dedicated node