Base Command Manager / Bright Cluster Manager Release Notes

Release notes for Bright 7.0-10

== cmdaemon ==

* Changed: Force all monitoring tables to MyISAM for improved performance. To manually convert the tables: On both head nodes stop cmd, run mysql -p cmdaemon_mon and execute 'ALTER TABLE MonData ENGINE=MyISAM; ALTER TABLE MonMetaData ENGINE=MyISAM; ALTER TABLE InfoMessages ENGINE=MyISAM; SHOW TABLE STATUS;' This could take a few minutes.
* Changed: mysql healthcheck to report FAIL if monitoring database tables are not MyISAM
* Added: failoverGroupMakeActive command to pythoncm bindings
* Added: /cm/local/apps/cmd/scripts/correct.monitoring.database.consolidators.py script to remove large number of unneeded NULL values in the monitoring database. This script should (but doesn't need) to be ran once for clusters installed with a revision before 21075
* Added: AdvancedConfig: AllowQuorumOnExternalNetworks=1 to allow for external networks to be used for HA quorum
* Added: check to prevent rare crash in UCS object XML conversion
* Added: extra check for hostname resolve, this can cause a crash in rare cases
* Added: check to prevent nodes from being added to multiple failover groups
* Added: extra metrics to get information on cmd mirroring of the monitoring database
* Added: advanced config flags: SampleOnDemandTimeout=200, SampleOnDemandThreads=25(PEThreads) to have better control over prejob checker
* Improved: cmha dbreclone for mysql >= 5.5
* Improved: do not force reinitialize of metric collection script if new metrics report "no data"
* Improved: propagate sampingMethod to all derived metrics when changing the prototype
* Improved: interfaces healthcheck now uses ip, instead of ifconfig
* Improved: When autosigning is off, issuing a certificate request defaulted in a certificate for admin instead of node profile. The default profile type is now determined by the request type.
* Fixed: infinite recursion in smart healthcheck for 3ware controllers if smart command does not work properly
* Fixed: in some cases, incorrect update of cmdaemon's cache of slurm jobqueues
* Fixed: removal of monitoring repository data when metric/healthcheck is deleted
* Fixed: typo in cmdaemon logrotate for sles
* Fixed: job prolog: when one of the prejob healthchecks failed, prolog still reported the node health is ok.
* Fixed: MIC healthchecks

== manuals ==

* Added: smtp relayhost configuration

== buildmaster ==

* Added: cm-hadoop-deployment-manual to default packages

== cluster-tools ==

* Hadoop: support for Cloudera CDH 4.7.0
* Hadoop: support for Pig project (cm-pig-setup)
* Hadoop: support for Cloudera CDH 5.0.2

== cm-openssl ==

* Update to version 1.0.1h

== cmgui ==

* Hadoop: add support for Cloudera CDH 4.7.0
* Hadoop: add support for Cloudera CDH 5.0.2
* Added: metric/healthcheck filter query
* Improved: allow devices inside a chassis to be placed outside the chassis in rack view
* Improved: refresh of rackview when chassis layout property changes
* Fixed: in some cases an issue with saving rack setup
* Fixed: eventviewer could be empty when using a 6.1 cmgui after a 7.0 cmgui

== mlnx-ofed-v2 ==

* Update from version 2.1-1.0.6 to 2.2-1.0.1.