When users hit 0 capacity issues, currently the only way for them to detect that is by checking the logs. This PR will leave the definition of the partition in azure.conf, but will comment it out with a comment that this particular partition/nodearray/vm_size has 0 capacity.
For example, I have 0 capacity for the ND40 here, this is what azslurm partitions writes out (and azslurm scale would put into /etc/slurm/azure.conf)
# The following partition has no capacity! htc - htc - Standard_ND40rs_v2
# PartitionName=htc Nodes= Default=NO DefMemPerCPU=16343 MaxTime=INFINITE State=UP
# Nodename=[] Feature=cloud STATE=CLOUD CPUs=40 ThreadsPerCore=1 RealMemory=653721 Gres=gpu:8
When users hit 0 capacity issues, currently the only way for them to detect that is by checking the logs. This PR will leave the definition of the partition in azure.conf, but will comment it out with a comment that this particular partition/nodearray/vm_size has 0 capacity.
For example, I have 0 capacity for the ND40 here, this is what azslurm partitions writes out (and azslurm scale would put into /etc/slurm/azure.conf)
# The following partition has no capacity! htc - htc - Standard_ND40rs_v2 # PartitionName=htc Nodes= Default=NO DefMemPerCPU=16343 MaxTime=INFINITE State=UP # Nodename=[] Feature=cloud STATE=CLOUD CPUs=40 ThreadsPerCore=1 RealMemory=653721 Gres=gpu:8