Base Command Manager / Bright Cluster Manager Release Notes

Release notes for Bright 8.0-6

== General ==

- Improvements

* Upgraded OpenStack Neutron and openvSwich packages

- Fixed Issues

* An issue with DHCP server support for InfiniBand (IPoIB) requests on Ubuntu

== cmdaemon ==

- Improvements

* Adv. conf. option NodeInfoPath to write files with category/image information for the node
* New GlobalConfig option GenerateLDAPCertificate to allow disabling of LDAP certificate generation
* An issue with the file path defined in CMDaemon for service shorewall on Ubuntu
* Improved page count for pagination in the user portal when custom filters are enabled
* Show the response from slurm commands in the user portal
* Improved filtering for jobs by date in the user portal
* Use "JobArrayID_TaskId" notation for slurm Job arrays in the user portal, which matches the notation used by slurm
* Improved sorting of jobs in the user portal

- Fixed Issues

* Improved handling of large sets of Azure nodes, which previously could cause high load averages and nodes' states flapping
* An issue with powering on Azure cloud nodes when it is performed immediately after powering them off
* An issue with starting CMDaemon due to Mylsql types changes in rhel7u4
* NFS OpenStack back-end combined with cloud bursting led to long provisioning time of cloud director
* sample_ipmi script --initialzize doesn't print all lines, which can result in an infinite initialize
* In some cases, an issue with managers getting out of date on the compute nodes
* Removed ipmi0 device from dev.init directory in node-installer root, the device is created automatically with the correct major number when ipmi kernel modules are being loaded.
* In some cases, CMDaemon crash while removing several AWS cloud nodes
* Cookies not parsed properly if both name and cm-login-token are provided
* Incorrect name for subnetmanager on Ubuntu nodes
* The default gateway was configured to its own IP for alias interface on Ubuntu16.04 clusters
* An issue with nodeRange field length, which in some cases could cause cm-scale to fail
* An issue with LDAP replication in HA setup when adding users while secondary head node is active
* An issue with /etc/hosts file generation in DirectConnect/ExpressRoute setups
* Removing a cloud provider could cause other cloud providers to become unreachable

== cluster-tools ==

- Improvements

* Improved examples in the help text for cm-hadoop-user
* Support for Apache Kafka 0.11.0.1
* Support for Cloudera CDH 5.12.1
* Support for Hortonworks HDP 2.6.2.0
* Support for Jolokia 1.3.7
* Support for Cloudera CDH 5.11.2
* Support for Apache Hadoop 2.7.4

- Fixed Issues

* Possible race condition when setting up docker

== Bright View ==

- Improvements

* Support for Cloudera CDH 5.12.1
* Support for Hortonworks HDP 2.6.2.0
* Support for Jolokia 1.3.7 in Bright View
* Support for Cloudera CDH 5.11.2
* Support for Apache Hadoop 2.7.4

== buildmaster ==

- Fixed Issues

* Include srvadmin-idracadm7 on the ISO

== cm-cluster-extension ==

- Fixed Issues

* An issue with naming in CMDaemon the VPC for cluster extension when deploying to a pre-existing VPC

== cm-create-image ==

- Fixed Issues

* Install cm-config-dracut-slave, cm-config-nfsclient when using -m option

== cm-openstack-setup ==

- Fixed Issues

* Improved detection of controller nodes' reboot status

== cm-scale ==

- New Features

* Allow job to be mapped to nodes that are in the job's queue
* Do not stop a cloud node if defined period of run time (WholeTime parameter) is not exceeded

== cmsh ==

- Fixed Issues

* Device list in cmsh may not show all devices

== cod ==

- Improvements

* Improved sort order for "image list", now accounting for Bright versions

- Fixed Issues

* cluster list may fail with datetime exception
* Added parameter pre-validation to avoid ending up with a partially created cluster on AWS
* Improved handling of COD_PREFIX environment variable.
* cluster-on-demand-openstack cluster list --columns now respects the specified order of columns

== monitoring ==

- Improvements

* Increased timeout for the defaultgateway healthcheck
* Added cmsh command to list enum values

- Fixed Issues

* No data sampled for Openstack project metrics.
* Monitoring did not account for IdleThreshold when set in cmd.conf
* Dynamic docker mount points were automatically monitored, now they are excluded
* Email action reports incorrect producer name
* JSON serialized measurable definition not parsing cumulative properly
* Handle large repository indexes properly, if storage index gets corrupted
* In some cases, cmdaemon crash if a monitoring index contains bad data
* In some cases, trigger history may get cleared after actions, resulting in new events / actions
* In some cases, CMDaemon crash on monitoring data check if a monitoring index contains bad data
* An issue which caused the defaultgateway healthcheck to report UNKNOWN instead of FAIL
* Improved date/time format parsing for cmsh dumpmonitoringdata
* An issue with diskspace healthcheck which caused it to not report FAIL
* dumpmonitoringdata did not allow for 4th argument in node device mode
* Changing a measurable cumumative not picked up by the monitoring system
* Do not include filtered out metrics for alterlevel / device status