Base Command Manager / Bright Cluster Manager Release Notes

Release notes for Bright 8.0-7

== cmdaemon ==

- Fixed Issues

* Non forwarded events can cause CMDaemon managers to go out of date
* An issue with several subsequent "power on" operations when they performed on the same Azure cloud node
* CMDaemon crash when run with -M command line option
* A service status in CMDaemon may be reset on CMDaemon restarts if those happen within a short time of each other
* An issue with bulk terminating large amount of Azure VMs, which may result in some of the resources not being deleted
* In some cases, NodeGroup DB not properly updated after a Node is removed
* NodeGroupManager can get out of sync on cloud directors
* An issue with stopping shorewall on RHEL6
* An issue which in some rare cases can cause CMDaemon to crash when removing cluster extension
* An issue with billing services signature mismatch

== cluster-tools ==

- Improvements

* Support for Apache Cassandra 3.11.1
* Support for Cloudera CDH 5.13.0
* Support for Alluxio 1.6.0
* Support for Apache Zeppelin 0.7.3
* Support for Apache Ignite 2.2.0

- Fixed Issues

* Hortonworks HBase 2.6.x cannot start for Hadoop on GPFS

== ceph ==

- Fixed Issues

* make cm-ceph-setup work with NVMe drives

== cm-clone-install ==

- Fixed Issues

* Slow network I/O transfer rate when booting UEFI compute nodes using TFTP

== cm-openstack-setup ==

- Fixed Issues

* Improved reliability of OpenStack deployment on slow clusters

== cm-scale ==

- Fixed Issues

* An issue with selecting a storage account for Azure nodes, which can make them slower
* In some cases, an issue with start/stop of a template node
* cm-scale does not check cloud provider default flavor if it is not set in the cloud node
* cm-scale can now get the CPU cores number from the WLM client role slots, if defined
* An issue with cm-scale stopping a node if SGE/UGE job is running and the node name exceeds 30 symbols

== cmsh ==

- Fixed Issues

* cmsh crash when certain commands are used in Puppet roles
* An issue with listcertificates --hiderevoked output
* An issue with displaying inherited values for azure cloud settings
* Rare cmsh crash when powering on Azure nodes

== cluster-on-demand ==

- Fixed Issues

* An issue with Azure head node becoming externally inaccessible after extending the cluster to a VNET in the same resource group
* An issue with instance type validation for cluster-on-demand in AWS
* An issue with Elastic IP assignment when creating multiple clusters in AWS

== monitoring ==

- Fixed Issues

* JobQueue monitoring triggers do not display the queue name, but the unique key
* Table with new-line strings not always properly indented
* Repository ID can get reused resulting in incorrect values for the latest health data
* In some cases, CMDaemon crash after samplenow
* cmsh JobQueue mode measurable command does not work with argument
* veth interfaces are now excluded from monitoring by default

== node-installer-slave ==

- Fixed Issues

* Slow network I/O transfer rate when booting UEFI compute nodes using TFTP