Base Command Manager / Bright Cluster Manager Release Notes
Release notes for Bright 9.1-6
== General ==
- Improvements
* Update cm-openssl to 1.1.1k
* Update Kubernetes to version 1.18.15
== cmdaemon ==
- Improvements
* Enable Ansible monitoring data producer modules
* Replace space and other special characters in category names when generating the tftpboot configuration
* Do ICMP ping in batches of 256 by default, to prevent socket buffer filling up
* Added boot node information to booting state
* Make the Lustre health check compatible with recent versions of Lustre
* Allow iblinkinfo.py to be executed from a different node (ibswitch.userdefined1 )
- Fixed Issues
* An issue with determining the home path for users in external-user-cert script
* In some cases, an issue with determining the correct user home directories for storing kube config file and certificates
* An issue with Ansible collection monitoring_data_producer_single_line_health_check_script's parameter startupDelay default value type
* Rare crash when performing PDU power operations
* Power delay for a 2nd node is not adhered to
* An issue with Unicode decode in the rogueprocess health check
* Support for windows-based usernames containing backslashes when creating Kubernetes resources
* An issue where multiple sessions from the same edge node are created on the head node
* In some cases, ntp fails on edge nodes due to a delayed named update
* An issue with updating a bond interface when renaming the underlying physical interface in cmsh
* An issue with writing all interfaces' MACs in the dhcpd config file, which can result in a boot loop in the case of a bond interface
* An issue with monitoring data for VMware nodes
* IBSwitch environment not passed to the switch info script
* An issue with Slurm slots autodetection when slots=0 in slurmclient role
* Possible issue with a failover group not becoming active again when all nodes are rebooted at the same time
* An issue where setting a primary group for a user also adds the user as a member of the group
== node-installer ==
- Improvements
* Possible crash of the node-installer when switch ports aren't correct and cannot be automatically detected
* Allow for updating the kernel parameters for nodes with installbootrecord=yes
- Fixed Issues
* Added BOOTIF information to kernel cmdline when booting with grub, which improves the boot interface detection by the node-installer
* Added fips dracut module to createramdisk for rhel and centos when available, to support fips in the original ramdisk
== cluster-tools ==
- Fixed Issues
* An issue with Slurm DB storage user credentials when Slurm is setup after head node HA has been setup
== Bright View ==
- New Features
* Allow different categories to be specified for edge director and edge nodes
- Fixed Issues
* In some cases, an issue with deleting a user
== cm-bios-tools ==
- Fixed Issues
* An issue with capitalization of some vendor_ids
== cm-cluster-extension ==
- Fixed Issues
* Calls to old boto library that has issues when a proxy is used have been changed to use newer boto3 calls.
== cm-scale ==
- Fixed Issues
* Possible crash when the region is not set in the cloud node cloudsettings
* An issue when mix_locations=false
== cm-tftpboot ==
- Fixed Issues
* Added BOOTIF information to kernel cmdline when booting with grub, which improves the boot interface detection by the node-installer
== cm-wlm-setup ==
- Fixed Issues
* In some cases, an issue with creating pbs.service file on GPU nodes
== cmsh ==
- Fixed Issues
* Some values in cmsh are expressed in B instead of B/s
* Postfix relayhost with port number writes an invalid main.cf
* cmsh does not allow renaming a job queue if the name is already used in a different WLM
* cmsh allows to select jobqueue from different WLM cluster
== cuda-dcgm ==
- Fixed Issues
* An issue with DCGM python bindings linking with DCGM libraries
== ml ==
- New Features
* Updated cm-cudnn8.1-* packages to v8.1.1.33
== pythoncm ==
- Improvements
* Added an example for job monitoring data dump
* Added image/fspart lock functions