You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi! I was looking to set up a funcx endpoint, where the funcx-endpoint start X command would be run by the flux broker, thus giving it access to an entire cluster in Kubernetes. This is what I was hoping to use for my config:
fromfuncx_endpoint.endpoint.utils.configimportConfigfromfuncx_endpoint.executorsimportFluxExecutorimportos# Retrieve settings from the environmentworking_dir=os.environ.get('FUNCX_WORKDIR')
# The executor will be run under the broker, with access to all its resourcesexecutor=FluxExecutor(working_dir=working_dir)
executor.launch_cmd="{flux} submit {python} {manager} {protocol} {hostname} {port}"config=Config(executors=[executor])
# For now, visible_to must be a list of URNs for globus auth users or groups, e.g.:# urn:globus:auth:identity:{user_uuid}# urn:globus:groups:id:{group_uuid}meta= {
"name": "default",
"description": "The Flux Operator FuncX Endpoint",
"organization": "Lawrence Livermore National Laboratory",
"department": "The Harry Potter Closet",
"public": False,
"visible_to": [],
}
To give some context, the endpoint would already be started in a flux instance with an allocated set of resources, so we would just need to receive a task and do flux submit via the Python sdk.
Thank you!
The text was updated successfully, but these errors were encountered:
Hi! I was looking to set up a funcx endpoint, where the
funcx-endpoint start X
command would be run by the flux broker, thus giving it access to an entire cluster in Kubernetes. This is what I was hoping to use for my config:But I was looking here: https://github.com/funcx-faas/funcX/blob/main/funcx_endpoint/funcx_endpoint/executors/__init__.py and I don't see it as an option. Is this something we could potentially add?
Note that it's already an executor in parsl (alongside the high throughput - why I am hoping we can add it here!)
https://github.com/Parsl/parsl/blob/7a235a4f5e285bc8deb7b57726f01cea001f4c89/parsl/executors/flux/executor.py#L1
To give some context, the endpoint would already be started in a flux instance with an allocated set of resources, so we would just need to receive a task and do
flux submit
via the Python sdk.Thank you!
The text was updated successfully, but these errors were encountered: