Share via

CycleCloud version 7.9.1

  • Article

The patch release improves the BeeGFS support.

New Features:

  • BeeGFS cluster type upgraded to enable NVMe SSDs on Ls_v2 VMs.
  • Stopping a BeeGFS cluster with managed disks is now supported.

Resolved Issues:

  • PBSpro nodes would occasionally fail initial converge due to hostname mismatch.
  • Cluster names containing spaces and other symbols would cause Jetpack installation failures.
  • Nodes that were deallocated would incorrectly show "Awaiting system boot-up" after being re-allocated.
  • Some cluster types would not automatically include inferred GridEngine and HTCondor projects.
  • Terminating and restarting Slurm clusters did not work properly when installing Slurm on a persistent disk
  • Slurm jobs failed to run due to RealMemory being set too high for some VM sizes
  • The "Host unreachable" message in the UI was not cleared properly when network connectivity is restored
  • The displayed node hostnames did not always match the VM hostnames
  • The HTTP Proxy settings form did not properly show the HTTPS proxy host
  • An unexpected error while collecting pricing data would cause Azure monitoring to fail
  • Dynamically created node arrays did not inherit the additional cluster-init specs of their parent
  • The BeeGFS client install was not working for HPC images