Base Command Manager / Bright Cluster Manager Release Notes
Release notes for Bright 6.1-8 (released 2013-08-10)
== cmdaemon ==
* Added SLES11sp3 support
* Move SSL error messages to debug log level
* Added ipmi identify token to the readonly profile
* Updated ilo_power.pl script, for updated HP iLO configuration script in node-installer package.
* Added: Ctrl-C handler to cmsh foreach loops
* Added: support to only listen on specific interfaces. AdvancedConfig: CMDaemonListenOnInterfaces
* Improved: allow -n, -r, -c options in cmsh dumphealthdata
* Improved: give a warning when adding a user with id < 500
* Fixed: objects not being marked as modified in cmsh after using swap command
* Fixed: store and parse lockDownDhcpd property of Network in xml
* Fixed: AT commands being sent to ttyMICx when using minicom via CMGUI
* Fixed: fill in new properties of LSF/openlava job
* Fixed: pythoncm eventhandler, so sessions are disconnected prematurely
* Fixed: GPU healthcheck includes.
* Fixed: installMode and nextBootInstallMode not being added to xml dump and init
* Fixed: VPC support in /etc/hosts and named
* Fixed: billing service not working when a proxy is set
* Fixed: ethernetswitch overview: mac addresses being returned twice on the first overview call
* Fixed: custom private IPs for VPC instances.
* Fixed: crash in namedservice when adding > 16 networks
* Fixed: ntp healthcheck: changed the way to determine if ntp daemon is running, previous method could cause time outs.
* Fixed: rare backup ping crash which may happen when an IP is null and a MIC is down
* Fixed: Reorganize metric collection order for ethernetswitches: do SwitchCPU first such that we do not measure the CPU load of the snmp calls done by the monitoring itself
* Fixed: possible small programrunner memory leak
* Fixed: issue with user cloud files transfer when one of $HOME directories is a link
* Fixed: prefer-passive IP and services being stopped on the passive during a hard failover when the machine becomes active
* Fixed: infomessage parsing from collection scripts
* Fixed: deadlock in pythoncm object synchronization
* Fixed: prolog.clouds in some cases not determining the cloud director
== node-installer ==
* Updated HP iLO locfg.pl script from version 3.00 to 4.10.
== node-installer-slave ==
* Update kernel for rescue environment for SLES11, from sp2 to sp3.
* create empty clone install log for ttys spawning
* cm-clone-install: does not rely on build-config from the head node anymore
== manuals ==
* updated: MPI (3) and OpenMP use
* added: compiler equivalents table
* added: drain dialog
* updated: PFX certificates are no longer used
* added: cm-dist-limit
== buildmaster ==
* Update PBS Professional logo, pbssales email address, include registered symbol with PBS Professional
* Install cm-openmpi-extra by default
* Added openvpn configuration files to grab exclude lists. This prevents openvpn configuration problems after doing a grab from a cloud node.
* Added SLES11sp3.
* Added bnx2x to default software image module list
* Update hwvendor package selection for DTK4.3 updates from Dell
== buildmatic-common ==
* Added SLES11sp3
== cciss_vol_status ==
* Update from version 1.10 to 1.11.
== cluster-tools ==
* Fixed: openlavaserver role not being unassigned when wlm-setup is used to disable openlava
* cmha-setup: honor the IP that is set for additional failover interfaces in the failover config in express mode
* Fixed cm-mic-setup when number of nodes equals number of pre-paid nodes
* cm-create-image: improvements to yum install error detection when exit code is non-zero due to post scriptlet failures in some rpms
* Added SLES11sp3
* cloud-check utility: added the "skipnet" and "onlynet" arguments
== cm-config-inittab ==
* Added SLES11sp3 to welcome script.
== cm-config-intelcompliance-master ==
* Intel cluster checker version updated from 2.0 to 2.1.
== cm-config-intelcompliance-slave ==
* Intel cluster checker version updated from 2.0 to 2.1.
== cm-config-minicom ==
* Set the modem init string to be empty. We use minicom to connect to MIC consoles, which don't need the init strings.
* Also cleared hang-up and reset strings to prevent sending those the MIC's.
== cm-config-motd ==
* Added SLES11sp3
== cm-config-mysql ==
* Added mysql 5.5 configuration for SLES11sp3.
== cm-config-upstart ==
* Added SLES11sp3 to welcome script.
== cm-config-yum ==
* fixed repo path in cm-extra.repo.oel5
== cmgui ==
* Merge JSON errors into one popup to prevent flooding of the desktop
* Added: support for images on local/shared disk in the provisioning role
* Fixed: cloud provider refresh jitter when finishing the cloud setup wizard
* Fixed: parsing diskless disklayout xml
* Fixed: rare case in which items are added twice to the resource tree
* Fixed: threshold rule wizard
* Fixed: MIC wizard mic0 IP overview (add base address)
* Fixed: exporting objects to other clusters
== cmsub ==
* Fixed: pbspro stdout/err files transfer when '#PBS -o/-e' are specified
* Fixed: use correct include path
* Fixed: only add 10mb resource limitation to sync jobs
== conman ==
* Fixed cm-conman init script, to let conman daemon re-read (reload) the config file with the init script.
== cuda-driver ==
* Driver update from 319.32 to 319.37.
* Added nvidia-healthmon to package.
== cuda50 ==
* Update of TDK from version 3.304.5 to 5.319.43.
== finalise-base ==
* Added creation of ecdsa ssh host key for SLES11sp3 software images.
== intel-cluster-checker ==
* Updated from 2.0 to 2.1
Due to updated XML libraries on SLES11sp3 clusters, the Intel Cluster Checker fails to run, because it can not read the XML configuration file. This issue is reported, an updated package will be made available in our yum/zypper repositories, as soon as a fix is available.
== intel-compilers-2013 ==
* Update from 2013-4 to 2013-5.
== intel-mic ==
* Upgrade to revision 16
== intel-mpi ==
* Updated from 4.1.0.030 to 4.1.1.036
== intel-tbb ==
* Update from 41_20130314oss to 41_20130613oss.
== node-installer-nfsroot ==
* Updated SLES11 packages to sp3.
== openmpi ==
* Move openmpi tm components to cm-openmpi-extra package (fixes PBS Pro issues with openmpi)
== torque ==
* Fixed: /etc/sysconfig/{torque_server,torque_sched,trqauthd} not being used in the init.d scripts
== verify-license ==
* Use /cm/local/apps/cmd/etc/cert.{pem,key} if not specified