Azure / Azure/cyclecloud-slurm
Easiest way to increase the number of nodes in an existing partition?
- Dominant language
- Python
- Stars
- 84
- Forks
- 56
- Avg merge
- 3d 17h
- Merged PRs (30d)
- 1
Description
Hi,
I have set up a cyclecloud cluster (after quite a bit of faffing, hence my hesitance to do anything that might require creating a new cluster) and now want to increase the number of available nodes in a specific partition.
What is the advisable way to do this?
My understanding is that I could edit the /etc/slrum/azure.conf file on the schedular node:
```
PartitionName=hpc Nodes=slurm-lab-cluster-subnet-hpc-[1-16] Default=YES DefMemPerCPU=6809 MaxTime=INFINITE State=UP
Nodename=slurm-lab-cluster-subnet-hpc-[1-16] Feature=cloud STATE=CLOUD CPUs=4 ThreadsPerCore=1 RealMemory=27238 Gres=gpu:1
```
updated to:
```
PartitionName=hpc Nodes=slurm-lab-cluster-subnet-hpc-[1-64] Default=YES DefMemPerCPU=6809 MaxTime=INFINITE State=UP
Nodename=slurm-lab-cluster-subnet-hpc-[1-64] Feature=cloud STATE=CLOUD CPUs=4 ThreadsPerCore=1 RealMemory=27238 Gres=gpu:1
```
I would then restart the slurmctld and slurmd daemons.
Alternatively, I wonder if I can update the settings on the Cyclecloud interface, save and apply the settings and then run azslurm scale CLI script on the schedular? My hesitancy with this is losing all the modifications I had to make to the slurm confs (including but not limited to setting up a slurm accounting database) to get the cluster properly running in the first case.
Any advice would be appreciated,
Thanks,
Noah
Contributor guide
No contributing guide indexed for this repository
Assessment
This issue has not been assessed yet.