Base Command Manager / Bright Cluster Manager Release Notes
Release notes for Bright 8.1-3
== General ==
- New Features
* Allow cloud nodes, when configured, to continue booting using kernel/initrd embedded on the node-installer image, rather than do standard update of kernel/initrd
- Fixed Issues
* Portal profile missing plot privileges, which can cause userportal charts to not work
* Ubuntu OS flavor is reported incorrectly in sysinfo
* Improved tokens and services update/management for profiles
* Ensure event unique keys are incremental as received by the client
* An issue with grouping by users for accounting & reporting default query
* Extra entry in device status DB table can be left behind by cmd -i
* Node installer certificate not issued if requested while cmd is starting
* In some cases, cloned compute nodes remain in UNKNOWN status after manual failover
* Failed health checks sometimes not reported in device status
* An issue with cm-diagnose monitoring collection commands
* cm-diagnose not collecting cmdaemon core dumps on SLES12
* An issue with shorewall6 not starting due to incorrect entries to /etc/shorewall6/netmap in cluster-extension setup
== cmdaemon ==
- Fixed Issues
* Do not uncomment attributes from original Dell XML from iDRAC
* sysinfo showing GPUs as not supported
* For Kubernetes API server, use the node's management IP for advertise address instead of active master
* cmdaemon crash when issuing "device consolelog" cmsh command for AWS nodes
* An issue with DNS configuration generation in the case of AWS + DirectConnect + two subnets in a VPC
* Change docker healthchecks to check for unhealthy instead of "dead" (which is not always a sign of a bad health)
* In disk setup, allow for specifying a blockdev which is a symlink (like many /dev/disk/*), this can be useful to prevent random disk swapping issues
* Trigger udev again in the node-installer so that any rules in /cm/node-installer/etc/udev/rules.d get applied
* In some cases, cmdaemon crash if cloud directory IP is updated before cloud node has come UP properly
== Bright View ==
- New Features
* Docker wizard
* Added changelog page
* Added Bright View editing shell (BvShell)
* Save the grid state customizations in the user profile
- Improvements
* Virtual gamepad in the rack view
* Support for range selection of rows in grids
- Fixed Issues
* Improved sidebar menu refreshing algorithm.
== cm-cluster-extension ==
- Fixed Issues
* cm-cluster-extension --remove will now attempt to remove the cloud-director's category
== cm-scale ==
- Improvements
* Support slurm logical job feature requests
- Fixed Issues
* In some cases, cmsub jobs are being ignored by cm-scale
* An issue with reading drain state if more then one workload manager has been setup
* Assigning a category to a not completely cloned node
* cm-scale dynamic provider leaves one less node not terminated than what is specified
* cm-scale now performs drain and poweroff operations on different iterations, making draining more reliable
== cmsh ==
- New Features
* ssh to compute nodes from within cmsh
- Fixed Issues
* Name range collapse format .. not taking into account the number of digits
== cod ==
- Fixed Issues
* COD now uses a COD_ prefixed environment variables (except for standard cloud credentials, like AWS_USERNAME, etc). This avoids conflicts for common names, like VERSION or IMAGE
* COD-OS cluster select with both names/patterns and version/distro will do AND instead of OR.
== monitoring ==
- New Features
* On-demand-only measurables are now saved in the database
* PreJob healthchecks are now visible in latesthealthdata
* Added --csv option to monitoring tool to import raw data
- Improvements
* Reduce the amount of monitoring storage logging on compute nodes
* Extend pythoncm monitoring example to use monitoring state information
* Added flags to hide disabled / filtered measurables in latest monitoring data
* Custom environment variables can now be set when running "samplenow"
* Support for dynamic monitoring resource for category and overlay
* Speed up sample now for large number of nodes
- Fixed Issues
* Enum value cache needs to be case insensitive for measurable / parameter and key values
* Trigger expressions need to be case insensitive
* cmsh monitoringdrop help not shown
* Usercount monitoring data producer counts all users
* Minimize the number of evaluated monitoring trigger expressions
* Monitoring "during" action not always executed
* Sample now of alertlevel doesn't translate measurable keys
* Prejob check not sampling if valid devices were not specified
* Prejob check on PBSPro uses FQDN instead of hostname
* In some cases, crash in prejob sampler
* An issue with cleaning up monitoring repositories after jobs are dropped because of their age
== wlm-setup ==
- Fixed Issues
* UGE setup on CentOS 6 reports failure