Base Command Manager / Bright Cluster Manager Release Notes
Release notes for Bright 8.0-5
== cmdaemon ==
- New Features
* Allow per-device specific username for internal cmdaemon ssh sessions
- Improvements
* Cluster extension to Azure: boot diagnostics for VMs are now removed along with the VM when it's being terminated
* OpenStack monitoring
- Fixed Issues
* cmdaemon not taking into account that kernel modules can be compressed
* The interfaces healthcheck shows false alarms for bridge interfaces with an additional /32
* Incorrect status for PBSpro jobs in the user portal when the jobs stop running
* Glance with filesystem backend store: an issue with glance-api.conf, which may results in the images stored by default in /var/lib/glance/images. Users should check the actual location for the images is set in OpenStack::ImageApi role prior to upgrading
== cluster-tools ==
- New Features
* Spark: always include Hadoop native libraries in LD_LIBRARY_PATH
- Improvements
* More robust interaction with ZooKeeper servers during validation
* Support for Apache Flink 1.3.2
* Support for Apache Storm 1.1.1
* Support for RDMA-Hadoop 2.x 1.2.0
* Support for Apache Ignite 2.1.0
* Support for Apache Drill 1.11.0
* Support for Apache Hadoop 2.8.1
* Support for Apache Hive 2.3.0
* Support for Cloudera CDH 5.9.3
* Support for Cloudera CDH 5.12.0
* Support for Apache Spark 2.2.0
- Fixed Issues
* Uninstall of Apache Drill does not remove configuration files in /etc on the head node
* Fixed sorting in Azure region list selection.
* SparkR job submissions fail with Spark 2.x
== Bright View ==
- Improvements
* Support for Apache Hadoop 2.8.1
* Support for Cloudera CDH 5.9.3
* Support for Apache Spark 2.2.0 in Bright View
* Support for Cloudera CDH 5.12.0
== cm-scale ==
- Fixed Issues
* An issue with disabling resource provider
* Allow cm-scale to terminate nodes that are not completely started
* Reduce cm-scale iteration time with about 30-50%
* Do not start cloud node when cloud director is not ready to serve the node
* An issue with resource provider priority usage
* An issue with cloud node cloning from the template node
== cmsh ==
- Improvements
* Allow selectors to be combined in CMSH commands
- Fixed Issues
* Incorrect/incomplete help for foreach command
* cm-scale: an issue with setting nodegroups in dynamic resource provider
== cod ==
- New Features
* cluster-on-demand-aws: added --ssh-public-key-path which can be used to embed a public SSH key on the COD cluster for the root user
- Improvements
* cluster-on-demand-azure now defaults to opening ports 80, 443 (landing page), 8081 (Bright view) on the COD cluster in azure
- Fixed Issues
* An issue with DNS configuration for cluster-on-demand in the cloud
* cluster-on-demand-azure: clusters can now have names, including upper case letters
* cluster-on-demand-aws: improved cluster creation progress reporting
* cluster-on-demand-aws: an issue with specifying head node AMI id when creating a new cluster
* cluster-on-demand-openstack: an issue with listing COD clusters when there are Heat stack floating_ip resources in a failed state
== wlm-setup ==
- New Features
* Support for GPUs in LSF