I found the same issue here Tasks fail when running an agent inside Notebook 2: Remote Agent · Issue #1204 · allegroai/clearml
# Copyright 2023 Google Inc.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Custom kernel launcher app to customize socket options."""
from ipykernel import kernelapp
import zmq
# We want to set the high water mark on *all* sockets to 0, as we don't want
# the backend dropping any messages. We want to set this before any calls to
# bind or connect.
#
# In principle we should override `init_sockets`, but it's hard to set options
# on the `zmq.Context` there without rewriting the entire method. Instead we
# settle for only setting this on `iopub`, as that's the most important for our
# use case.
class ColabKernelApp(kernelapp.IPKernelApp):
def init_iopub(self, context):
context.setsockopt(zmq.RCVHWM, 0)
context.setsockopt(zmq.SNDHWM, 0)
return super().init_iopub(context)
if __name__ == '__main__':
ColabKernelApp.launch_instance()
Oh!
I see this is using the colab as remote agent (i.e. to launch jobs on it),
[ColabKernelApp] CRITICAL | Bad config encountered during initialization: The 'kernel_class' trait of <main.ColabKernelApp object at 0x7fa41b29e5c0> instance must be a type, but 'google.colab._kernel.Kernel' could not be imported
Can you send the full log?
Hi @<1686547344096497664:profile|ContemplativeArcticwolf43>
In the 2nd 'Getting Started' tutorial,
Could you send a link to the specific notebook?
. But whenever a task is picked, it fails for the following
You mean after the Task.init
call?
Hi @<1523701205467926528:profile|AgitatedDove14> , thanks for the reply. This happens after I enqueue
the task in the queue where the colab agent is registered. Please see None to have a look at the outputs
the task is a clone of one previously executed on google colab too
It actually started executing your code, but it did not capture it correctly:
/root/.clearml/venvs-builds/3.10/bin/python -u /root/.clearml/venvs-builds/3.10/code/colab_kernel_launcher.py
Which I assume means the actual Task had bad code.
What do you have under the Task execution tab in the UI (the one you were launching, i.e. enqueueing )
Good news is that I've tried again running from a fresh colab instance (run initial task, clone, run agent, send cloned task to agent) and it seems to be working now