Base Command Manager / Bright Cluster Manager Release Notes

Release notes for Bright 7.0-19

== cmdaemon ==

* Added: pythoncm commands to get all latest (or for a time interval) monitoring data for all devices, and examples
* Added: support for spaces in /etc/export paths
* Added: adv. config option AutomaticMetricBlackList to black list automatically added metrics
* Added: adv. config option OpenStackWorkerProcesses to configure the number of nova API processes
* Added: adv. config option OpenStackInternalnetRouterIP to configure the IP of the Neutron router port on the internal network
* Improved: conman.conf file when using cimc0 interfaces
* Improved: a standalone compute node is now considered the active member of a node-HA, needed when services are marked runIf ACTIVE
* Fixed: UCSBiosBootDevGrp being sent during ucsset
* Fixed: in some cases, CMDaemon generating gender files that do not respect the HOSTLIST format
* Fixed: CONNECTED_MODE for IB interfaces which are members of a bond
* Fixed: Hadoop user HDFS access timing issue: wait for HDFS safe mode to finish
* Fixed: munge.key permissions in Xeon Phi
* Fixed: Hadoop topology.data file not being updated on all NameNodes
* Fixed: restart of Hadoop namenode service after topology files changes
* Fixed: closing a device left it in a perpetual opening state after head node cmd restart
* Fixed: dumpmetricdata not merging latestmetricdata element with the last saved element if the rates are identical
* Fixed: duplicate monitoring data issue when plotting constant value over a long period

== node-installer ==

* Improved: sreboot and shalt no longer require /tmp filesystem

== manuals ==

* Added: open --failbeforedown, --reset options
* Added: CookieExpiration (cmgui connection cookie expiration)
* Added: additional hadoop projects
* Added: NameNode configuration choices
* Updated: LDAP now supported in MPSS

== buildmaster ==

* Make high speed ethernet network configuration indicate 10GE and 40GE. The corresponding layer-3 network name is now hsethernet instead of 10genet
* install rpmlint, rpmdevtools and dependencies on redhat6 distros to be able to rebuild mellanox ofed drivers for intel-mic

== cluster-tools ==

* Hadoop: add support for Apache 2.5.0
* ucsmanage: hack for bad attribute string containing < from CIMC version 2.0(3d)
* Fixed: cmha-setup crash when viewing failover status while the secondary head node is booting

== cm-config-intelcompliance-master ==

* Removed dapl package from the requirement list.

== cm-config-intelcompliance-slave ==

* Removed dapl package from the requirement list.

== cmgui ==

* Hadoop: add support for Apache Hadoop 2.5.0
* Added: new user authentication types (introduced in MPSS 3.x) to MIC settings
* Fixed: cmgui->exit hanging on slow connections

== cuda-driver ==

* Update driver from version 331.89 to 340.29. Update gdk version from 331_62 to 340_29.

== cuda65 ==

* Added: cuda65 for Bright 7.0

== intel-mic ==

* Fixed: configuring ldap via micctrl utility
* Fixed: compiling MIC-enabled code with Intel compiler
* Fixed: ofed driver installation
* Fixed: ibp_server kernel module when Mellanox OFED is used

== intel-wired-ethernet-drivers ==

* Update ixgbe driver from 3.21.2 to 3.22.3 and ixgbevf from 2.14.2 to 2.15.3.
* Added: i40e and i40evf drivers. Update igb driver from 5.2.5 to 5.2.9.4.

== node-installer-nfsroot ==

* Added: rpm needed for configuring hardware RAID.

== pgi ==

* Update from 14.4 to 14.7.

== qlgc-ofed ==

* Update from version 7.2.2.0.8 to 7.3.0.0.26.

== sge ==

* Fixed: incorrect paths in cm-launcher and epilog (cleanup-resources) scripts

== torque ==

* Build TORQUE with MPSS 3.x support