This is an old revision of the document!
In order to boot the HPC VMs the system administrator must run the following commands
vm start thin-01 # start the vm thin-01 ping thin-01 # make sure that you get a response back before starting a different VM vm start thin-02 ping thin-02 #optionally, if you are not scared vm start thick-01 ping thick-01
Finally, after all the HPC VMs have booted, you may boot the login nodes
vm start ssh-01 ping ssh-01 vm start ssh-02 ping ssh-02
There might be other VMs for other users that you might need to run
vm start rshiny0 ping rshiny-0 vm start rshiny1 ping rshiny-1
here's how the output of vm list would look like
root@geonosis:~ # vm list NAME DATASTORE LOADER CPU MEMORY VNC AUTO STATE comp0 default uefi 32 32G - No Stopped dl-01 default uefi 2 8G - No Stopped dl-02 default uefi 2 8G - No Stopped dna0 default uefi 8 32G - No Stopped mitte-dev-01 default uefi 2 4G - No Stopped rshiny0 default uefi 4 16G 127.0.0.1:5900 No Running (90735) rshiny1 default uefi 4 16G 127.0.0.1:5901 No Running (6542) ssh-01 default uefi 2 4G - No Running (65113) ssh-02 default uefi 4 8G - No Running (29385) thick-01 default uefi 64 768G - No Locked (geonosis.local.abi.am) thin-01 default uefi 64 384G - No Running (93403) thin-02 default uefi 64 384G - No Running (69252)
Finally, you will need to update the SLURM configs.
ssh scontrol-01 -l ops # become root su -
attach to tmux session that's running SLURM
the following command will show the issues
sinfo -R
you can “idle” the nodes after they have been booted
scontrol update State=idle NodeName=thin-01
note that this command is idempotent and will run once between reboots (unless the state gets corrupted somehow)
finally run the fancy command to see what's happening
clear; (sinfo -R; sinfo -N -o "%15N %9P %8a %5c %10m %20C %10e" ; squeue -o "%.6i %.10P %.10j %.15u %.10t %.10M %.10D %.20R %.3C %.10m") | column -t
to detach from tmux in tmux, you need to run prefix + prefix, in this case: ctrl-b, ctrl-b, d
never ever shut down the tmux session in scontrol. tmux is running in the foreground. you may exit now