-
Notifications
You must be signed in to change notification settings - Fork 115
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[BUG]TaskGraph UI lacking "Add Nodes" menu item to start working #154
Comments
Could you look in the terminal where jupyter lab was launched from and share if you have any errors printed out? If one of the nodes fails to load, then the "Add Nodes" menu will not show up. Most likely it's a misconfiguration of one of the plugins. We should enhance Greenflow to skip faulting nodes and/or plugins, and print out exceptions explaining the cause. But currently, you have to look in the terminal for the errors to find the faulty node and/or plugin. In the error you shared
or PyPi:
If you cd into the "gQuant/plugins/gquant_plugin/notebooks" directory launch jupyter lab in that directory. If you cd into the "gQuant/plugins/gquant_plugin" and launch jupyter lab from this top level directory there's a "greenflowrc" file that has:
You can just mv/rm that greenflowrc. Alternatively start jupyter lab with a MODULEPATH environment. Example:
The UI will be slightly different from the animation, since the animations were made a while ago. But it will be similar enough to follow along. |
I refined my start dir/cmdl a bit according to your suggestion, and here is the background terminal log:
Seems the direct failure is I created the export RAPIDS_VERSION=21.06
export RAPIDS_CUDA_VER=11.2
conda install -y mamba -n base -c conda-forge
mamba install -y -c rapidsai -c nvidia -c conda-forge -c defaults \
rapids=$RAPIDS_VERSION cudatoolkit=$RAPIDS_CUDA_VER python=3.8 && \
conda clean --all -y
mamba install -y -c conda-forge -c defaults \
jupyterlab'>=3.0.0' jupyter-packaging'>=0.9.2' jupyterlab-system-monitor \
nodejs=12.4.0 python-graphviz pydot ruamel.yaml && \
conda clean --all -y && \
jlpm cache clean && \
jupyter lab clean
pip install bqplot==0.12.21 && \
jlpm cache clean && \
jupyter lab clean
# or install from PyPi https://pypi.org/project/greenflowlab/
pip install greenflowlab
# greenflowlab will install greenflow as well.
# plugins are here https://github.com/NVIDIA/fsi-samples/tree/main/gQuant/plugins
# cd to each directory and install
cd gQuant/plugins/gquant_plugin
pip install . I installed greenflowlab/greenflow from pip, but gquant_plugin from the locally checkout repo, not quite sure what combination should be more desirable? |
Update:
Though pip would wrongly upgrade numpy to 1.22.1 causing numba to fail, I managed to fool conda/mamba to downgrade numpy with Now I have good looking [Add Nodes] UI, but indicator_demo.ipynb still suffers the |
A lucky guess took me a step forward w.r.t. task_load_csv_data = {
TaskSpecSchema.task_id: "load_csv_data",
TaskSpecSchema.node_type: "CsvStockLoader",
TaskSpecSchema.conf: {"file": "../data/stock_price_hist.csv.gz"},
TaskSpecSchema.inputs: {}
} To: task_load_csv_data = {
TaskSpecSchema.task_id: "load_csv_data",
TaskSpecSchema.node_type: "CsvStockLoader",
TaskSpecSchema.conf: {"file": "../data/stock_price_hist.csv.gz"},
TaskSpecSchema.inputs: {},
TaskSpecSchema.module: 'greenflow_gquant_plugin.dataloader',
} Then it'll fail with So is this the way codeful graph nodes are supposed to be written? Should I report a bug against |
So looks like the ray API has changed. We tested in Summer June 2021 (hence rapids 21.06). There is a dependency for gquant plugin on ray:
Reinstall ray in your python environment like this:
Let's focus on "01_tutorial.ipynb" and I can help out with "indicator_demo.ipynb" after getting greenflow running. |
Sure, the UI works for me now w.r.t. notebooks. Shall we close this issue now, or leave it open to track an explicitly version requirement of
|
Good, I'm happy the UI works and you're able to see the plugins in the "Add Nodes" menu. We can leave this issue open. Once I get around to it I'll update the setup in gquant plugin and pin the ray version. Then I'll close this issue. Thanks for uncovering it. |
Describe the bug
The actual RMB click menu appears very different than the screenshot animation, especially there is no [Add Nodes] entry so no way to start working at all.
Steps/Code to reproduce bug
Open notebook https://github.com/NVIDIA/fsi-samples/blob/main/gQuant/plugins/gquant_plugin/notebooks/01_tutorial.ipynb
Try follow the steps demonstrated in the screenshot animation, but right mouse button click gives a menu UI like this:
Expected behavior
To have the same or equivalent UI to accomplish the steps.
Environment overview (please complete the following information)
Environment details
Please run and paste the output of the
/print_env.sh
script here, to gather any other relevant environment detailsN/A
Additional context
https://github.com/NVIDIA/fsi-samples/blob/main/gQuant/plugins/gquant_plugin/notebooks/cuIndicator/indicator_demo.ipynb would err out like this:
The text was updated successfully, but these errors were encountered: