Sometimes HyperQueue is not in charge of creating Slurm/PBS allocations, but it is already started in an environment with pre-allocated nodes, and the users then just need to deploy a HQ worker on each node. It would be nice if HQ could help with this, and support the most obvious use-case, deploying through SSH.
Sometimes HyperQueue is not in charge of creating Slurm/PBS allocations, but it is already started in an environment with pre-allocated nodes, and the users then just need to deploy a HQ worker on each node. It would be nice if HQ could help with this, and support the most obvious use-case, deploying through
SSH
.The command could look like this:
It would read the node file, and essentially run
hq worker start <worker-start-params>
on each node.