I tried to start ray using the following command (Ray 2.3.0):
ray start --head
The following dashboard error came up.
ERROR services.py:1169 -- Failed to start the dashboard , return code -11
I tried to run the command with the following option and still get the same error:
--include-dashboard=False
The dashboard.log has the following messages:
2023-03-07 11:59:07,915 INFO head.py:135 -- Dashboard head grpc address: 0.0.0.0:35849
2023-03-07 11:59:07,917 INFO head.py:239 -- Starting dashboard metrics server on port 44227
The dashboard.err is blank.
cc @sangcho for thoughts.
Which python version do you use?
I’m using Python 3.8.16. What does the code -11 mean?
It somehow related to creating an environment with tensorflow involving conda-forge, i.e.
conda create -n test tensorflow python=3.8 root -c conda-forge
and running ray in this environment will has above error, but
conda create -n test tensorflow python=3.8
and running ray in this environment will be fine
The reason I need to invoke “-c conda-forge” is to install ROOT with conda.
It’s most likely because one shouldn’t conda install tensorflow in conda-forged channel. There seems to be some inconsistency. Thanks for all the follow up. I will close this thread now