Base Command Manager / Bright Cluster Manager Release Notes
Release notes for Bright 7.2-28
== cmdaemon ==
* Changed: default cm-dell-manage script timeout to 90s
* Added: comment in dhcpd config writer so it is easier to detect VSMP primary nodes
* Added: roles mode for MIC node categories in cmsh
* Fixed: in some cases, ceph.conf file was not updated when Ceph properties were changed in cmdaemon
* Fixed: in rare cases, cmdaemon crash when setting/removing service states
* Fixed: pythoncm cluster.find(name, baseType) not working for numeric names
* Fixed: ngpus and gpus complex value set for SGE and UGE.
* Fixed: Spark master and historyserver not started when headnode is restarted
* Fixed: fspart is displayed as a number in softwareimage (and wrong for boot)
* Fixed: gpuhealth healthcheck now reports unknown if it can not determine the result of a known nvidia-healthmon sub-test.
* Fixed: changing the name of an interface of a bridge does not update the bridge interfaces-list
* Fixed: an issue with serving admin/user pdf through the userportal
* Fixed: failure to add groups longer that 15 characters
* Fixed: running commands on nodes in a specified --state does not account properly for CLOSED state
* Fixed: slurm/gres.conf not written on slurmrole->gpuunits change
* Fixed: an issue with starting slurmd when slurm configuration is frozen in cmd.conf
* Fixed: pythoncm cluster.reboot doesn't work
* Fixed: an issue with showing the submit, start, and finish time of completed Slurm jobs when checking job history
* Fixed: Allow users to connect with cmsh to cluster with special port
* Fixed: cmdaemon crash in CertManager::doMaintenance (std::_Rb_tree_increment)
* Fixed: compute nodes booting in UEFI mode fails when their category has a defaultgateway set to an IP address outside of the boot network
* Fixed: Upgrade 7.{0,1}/6.1 -> 7.{2?,3}: cgconfig service not added on nodes with slurmclient/torque_mom role
* Fixed: CMDaemon adds the MIC nodes to the license's node count
* Fixed: in some cases, cmdaemon did not generate /etc/mkinitrd_cm.conf for the software image
* Fixed: galera database directory not in the grabimage exclude list
* Fixed: diskspace healthcheck may fail because the output of df is wrongly parsed.
* Fixed: minicom doesn't work for MIC nodes
* Fixed: in some cases, an issue with setting connected/datagram IB mode for suse
* Fixed: Node fails to set its cmdaemonUrl due to validation of BMC information
* Fixed: CMDaemon ignores "NtpConfigStaticServers127" advanced config and adds the local clock to /etc/ntp.conf
* Fixed: crash when removing a node at the same time as requesting the synclog
* Fixed: Don't set MINICOM enviroment variable anymore for minicom command for MIC device
* Fixed: CMDaemon generates incorrect ifcfg- files which are members of bond (SLES12sp1)
== node-installer ==
* Added: option in node installer to automatically allow swapped switch ports
* Fixed: The node-installer fails if the the external interface is tagged on the internal interface while the internal interface is bridged
* Fixed: ramdisk creation for nodes fails on btrfs readonly filesystems
* Fixed: Boot interface as a bond member, with bond part of a bridge causes boot failure
== node-installer-slave ==
* Updated: for SLES11 clusters now the SLES11sp4 kernel is used as base for the rescue kernel and for SLES12 clusters now the SLES12sp1 kernel is used as base for the rescue kernel.
== admin-manual ==
http://support.brightcomputing.com/manuals/7.2/admin-manual.pdf
* Fixed: Maui patched to work for Torque beyond v4
== cloudbursting-manual ==
http://support.brightcomputing.com/manuals/7.2/cloudbursting-manual.pdf
* Added: snapshots
== hadoop-deployment-manual ==
http://support.brightcomputing.com/manuals/7.2/hadoop-deployment-manual.pdf
* Updated: version updates
== installation-manual ==
http://support.brightcomputing.com/manuals/7.2/installation-manual.pdf
* Updated: Intel Compiler Suite 2016
== cluster-tools ==
* Added: support for Accumulo 1.7.2
* Added: support for Hive 2.1.0
* Added: support for Spark 1.6.2
* Added: support for Apache Drill 1.7.0
* Added: support for Spark 2.0.0
* Added: support for Apache Zeppelin 0.6.1
* Added: support for Storm 1.0.2
* Added: support for Kafka 0.10.0.1
* Added: support for Apache Ignite 1.7.0
* Added: support for Apache Flink 1.1.1
* Added: support for Hortonworks 2.5.0
* Added: support for Apache 2.7.3
* Added: support for Zookeeper 3.4.9
* Added: support for Hbase 1.2.3
* Added: support for Accumulo 1.8.0
* Fixed: error in wlm-setup on rhel6 and sles11 systems when some 3rd party rpm creates /usr/lib/systemd/system directory
* Fixed: cm-hadoop-setup -u not cleaning /var/lib/kms
* Fixed: CM MIC setup, cm-mic-setup. Previously cm-mic-setup failed when trying to rename a alias interface when the alias tag was a non integer value. Now cm-mic-setup is also able to rename alias interfaces with a non integer value tag.
== cm-apache-hadoop ==
* Added: support for Hbase 1.2.2
* Added: support for Apache 2.7.3
* Added: support for Zookeeper 3.4.9
* Added: support for Hbase 1.2.3
== cm-apache-hadoop-extras ==
* Added: support for Accumulo 1.7.2
* Added: support for Accumulo 1.7.2
* Added: support for Hive 2.1.0
* Added: support for Spark 1.6.2
* Added: support for Apache Spark 2.0.0
* Added: support for Storm 1.0.2
* Added: support for Kafka 0.10.0.1
== cm-cloudera-hadoop ==
* Added: support for Cloudera 5.8.0
== cm-hortonworks-hadoop ==
* Added: support for Hortonworks 2.5.0
== cm-mtar ==
* Added: mtar (bsdtar) for Bright 7.2
== cm-scale-cluster ==
* Fixed: crash when ASSIGN_CATEGORY parameter is applied to template nodes that do not exist yet
* Fixed: in some cases cloud-director was being powered off when cloud nodes are still UP in the same region
* Fixed: changing node categories when node is terminated
* Fixed: assigning category to a terminated node
== cm-webportal ==
* Adds missing JSON API documentation link
== cmgui ==
* Added: support for Spark 1.6.2
* Added: support for Cloudera 5.8.0
* Added: support for Hbase 1.2.2
* Added: support for Hortonworks 2.3.6
* Added: support for Cloudera 5.7.2
* Added: support for Spark 2.0.0
* Added: support for Hortonworks 2.5.0
* Added: support for Apache 2.7.3
* Added: support for Zookeeper 3.4.9
* Added: support for Hbase 1.2.3
* Added: support for Hortonworks 2.4.3
* Added: support for Cloudera 5.7.3
* Added: support for Cloudera 5.5.5
* Improved: allow cmgui to run with bundled firefox
* Fixed: cmgui does not allow for setting DataNode for category
* Fixed: node creation wizard failes when switch names are too long
* Fixed: Drag and drop doesn't work anymore with firefox 51
* Fixed: drac power control not shown in cmgui
* Fixed: in some cases, an issue with duplicate serials preventing cmgui from connecting
== cuda-driver ==
* Updated: to version 352.99
* Updated: nvidia kernel driver to version 361.93.02.
== cuda80 ==
* Added: CUDA Toolkit 8.0 packages
== intel-compilers-2016 ==
* Updated: to version 2016 update 4.
* Fixed: previously compiler_xe eclipse files were in the wrong directory. Now the are placed in the correct directory.
== intel-mic ==
* Updated: to MPSS version 3.7.1.
* Updated: to MPSS 3.7
* Updated: MPSS to version 3.7.1.
== intel-tbb ==
* Updated: to version 2017_20160722oss.
== intel-wired-ethernet-drivers ==
* Updated: e1000e driver to 3.3.5
* Updated: e1000e driver to 3.3.4, i40e driver to 1.5.19, fm10k driver to 0.21.3.
== libpam_wlm ==
* Fixed: failure on sles11
== mlnx-ofed33 ==
* Added: mlnx-ofed33 for Bright 7.2
* Updated: to version 3.3-1.0.4.0.
== openstack-nova ==
* Updated to 12.0.1