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 am trying to get NVFlare (v2.0.15) running on the Summit machine at Oak Ridge National Lab (IBM Power9 / V100), and seem to be running into an issue with server startup. It's coming from protobuf (v3.20.1) and had wondered if you had heard or experienced this before? I'm not able to find any info on stackoverflow or github, otherwise.
bash-4.4$ jsrun -n1 ./poc/server/startup/start.sh
WORKSPACE set to /gpfs/alpine/stf018/proj-shared/jaellis2/privacy/examples/poc/server/startup/..
sub_start: WORKSPACE set to /gpfs/alpine/stf018/proj-shared/jaellis2/privacy/examples/poc/server/startup/..
sub_start: Prepare to train nvflare server
Traceback (most recent call last):
File "/ccs/home/jaellis2/.conda/envs/smartsim/lib/python3.8/runpy.py", line 194, in _run_module_as_main
return _run_code(code, main_globals, None,
File "/ccs/home/jaellis2/.conda/envs/smartsim/lib/python3.8/runpy.py", line 87, in _run_code
exec(code, run_globals)
File "/ccs/home/jaellis2/.conda/envs/smartsim/lib/python3.8/site-packages/nvflare/private/fed/app/server/server_train.py", line 27, in <module>
from nvflare.private.fed.app.fl_conf import FLServerStarterConfiger
File "/ccs/home/jaellis2/.conda/envs/smartsim/lib/python3.8/site-packages/nvflare/private/fed/app/fl_conf.py", line 24, in <module>
from nvflare.private.fed.client.base_client_deployer import BaseClientDeployer
File "/ccs/home/jaellis2/.conda/envs/smartsim/lib/python3.8/site-packages/nvflare/private/fed/client/base_client_deployer.py", line 18, in <module>
from nvflare.private.fed.client.fed_client import FederatedClient
File "/ccs/home/jaellis2/.conda/envs/smartsim/lib/python3.8/site-packages/nvflare/private/fed/client/fed_client.py", line 28, in <module>
from ..utils.numproto import proto_to_bytes
File "/ccs/home/jaellis2/.conda/envs/smartsim/lib/python3.8/site-packages/nvflare/private/fed/utils/numproto.py", line 21, in <module>
from nvflare.private.fed.protos.federated_pb2 import NDArray
File "/ccs/home/jaellis2/.conda/envs/smartsim/lib/python3.8/site-packages/nvflare/private/fed/protos/federated_pb2.py", line 19, in <module>
from google.protobuf import descriptor as _descriptor
File "/ccs/home/jaellis2/.conda/envs/smartsim/lib/python3.8/site-packages/google/protobuf/descriptor.py", line 47, in <module>
from google.protobuf.pyext import _message
TypeError: bases must be types
System is in trouble and unable to start the task!!!!!
I needed to follow these instructions to install NVFlare without tenseal: #130
Separately, I have NVFlare running on our x86 clusters just fine. I can give plenty more detail on my env if needed. Thanks for any help!
The text was updated successfully, but these errors were encountered:
@ellisja can you probably can check if the protobuf version used works on the PowerPC, as separate test, and if it doesn't, see which version it works with. then we can help to investigate. Its a bit difficult to investigate since we don't have PowerPC
Hi!
I am trying to get NVFlare (v2.0.15) running on the Summit machine at Oak Ridge National Lab (IBM Power9 / V100), and seem to be running into an issue with server startup. It's coming from protobuf (v3.20.1) and had wondered if you had heard or experienced this before? I'm not able to find any info on stackoverflow or github, otherwise.
I needed to follow these instructions to install NVFlare without tenseal: #130
Separately, I have NVFlare running on our x86 clusters just fine. I can give plenty more detail on my
env
if needed. Thanks for any help!The text was updated successfully, but these errors were encountered: