Base Command Manager / Bright Cluster Manager Release Notes
Release notes for Bright 8.2-22
== General ==
- Improvements
* cuda-driver: updated to version 450.51.06
* cuda11.0: updated to version 11.0 update 1
* Added CUDA 11.0 packages
* Updated cuda-driver to version 440.95.01
== cmdaemon ==
- New Features
* Do not create /etc/slurm/slurm.conf symlink if Slurm is frozen
- Improvements
* Added memory / swap total to partition metrics
* Added node based version information
* New advanced config flags to disable individual sysinfo collection
* Added an automatic retry for system info put, to overcome moments when the head node is too busy to process them
* Clean up orphaned pexec trackers from the database to prevent slow cmd stop
* Improved parallel executor bookkeeping so they don't timeout after a day
* Open files in /proc with read only
* Edge director /etc/host missing spaces between some host definitions
- Fixed Issues
* REST calls do not return full measurable name, :parameter missing
* /etc/localtime could not be frozen from cmd.conf
* dumpmonitoringdata --min argument returns the sum instead of the minimum values across entities
* Exporting and re-importing cmd configuration can cause loss of monitoring healthchecks
* Job filter crash when a job no longer exists
* Deadlock in dcgm when connections are slow or not established
* TotalGPUUtilization metric value remains 0
* Edge hash secrets did not always get saved in DB
* Escape # in the password for the freeipmi configuration file
* Monitoring reinitialize didn't work for edge nodes
* An issue with parsing of comma's in openssl subject fields of existing license in request-license
== node-installer ==
- Fixed Issues
* The node-installer attempts to always use the base partition BMC credentials even if it is overridden at the node/category level
== cmha-setup ==
- Fixed Issues
* fuser path setting for Ubuntu and SLES in dasumount.sh script
== cmsh ==
- Fixed Issues
* dumpmonitoringdata --min argument returns the sum instead of the minimum values across entities
== ml ==
- New Features
* Updated cm-horovod-* packages to v0.20.0
* Updated cm-xgboost-* packages to v1.2.0
* Updated cm-mxnet-* packages to v1.7.0
* Updated cm-gpytorch-* packages to v1.2.0
* Updated cm-fastai-* packages to v1.0.63
* Updated cm-cmake-* packages to v3.18.2
* Updated cm-chainer-* packages to v7.7.0
* Updated cm-pytorch-* packages to v1.6.0
* Updated cm-theano-* packages to v1.0.5
* Updated cm-opencv3-* packages to v3.4.11
* Updated cm-horovod-* packages to v0.19.5
* Updated cm-nccl2-* packages to v2.7.8
== pythoncm ==
- Improvements
* HTTP proxies were used by default
== slurm ==
- Fixed Issues
* An issue with Slurm power scripts