Error while trying to Launch Tensorboard in the cell of Jupyter Notebook

[worldpeaceaspirer]

Issue description

I'm trying to Launch Tensorboard from the Jupyter Notebook cell but it is resulting in error. My machine doesn't have a GPU and from the error message, I understand that Tensorboard is expecting GPU.

It is working fine in Google Colab.

Can't the Tensorboard be launched in the Local PC (Linux OS), which doesn't have a GPU, from a Jupyter Notebook cell?

Please find the reproducible code and the Error, below:

%load_ext tensorboard
!rm -rf logs

import tensorflow as tf
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense
import numpy as np

model = Sequential([Dense(units = 3, input_shape = [1]),
                   Dense(units = 1)])
model.compile(optimizer = 'sgd', loss = 'mse')

x = np.array([-1.0, 0, 1, 2, 3, 4.0], dtype = 'float')
y = np.array([-3, -1.0, 1, 3.0, 5.0, 7.0], dtype = 'float')
tensorboard_callback = tf.keras.callbacks.TensorBoard(log_dir="logs")

model.fit(x, y,epochs = 500, 
          batch_size = 2,
         verbose=1, callbacks = [tensorboard_callback])

%tensorboard --logdir logs

Error Message:

ERROR: Failed to launch TensorBoard (exited with 1).
Contents of stderr:
2021-05-21 18:31:51.972053: W tensorflow/stream_executor/platform/default/dso_loader.cc:60] Could not load dynamic library 'libcudart.so.11.0'; dlerror: libcudart.so.11.0: cannot open shared object file: No such file or directory
2021-05-21 18:31:51.972099: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.
Traceback (most recent call last):
  File "/home/mothukuru/anaconda3/bin/tensorboard", line 8, in <module>
    sys.exit(run_main())
  File "/home/mothukuru/anaconda3/lib/python3.7/site-packages/tensorboard/main.py", line 75, in run_main
    app.run(tensorboard.main, flags_parser=tensorboard.configure)
  File "/home/mothukuru/anaconda3/lib/python3.7/site-packages/absl/app.py", line 300, in run
    _run_main(main, args)
  File "/home/mothukuru/anaconda3/lib/python3.7/site-packages/absl/app.py", line 251, in _run_main
    sys.exit(main(argv))
  File "/home/mothukuru/anaconda3/lib/python3.7/site-packages/tensorboard/program.py", line 289, in main
    return runner(self.flags) or 0
  File "/home/mothukuru/anaconda3/lib/python3.7/site-packages/tensorboard/program.py", line 305, in _run_serve_subcommand
    server = self._make_server()
  File "/home/mothukuru/anaconda3/lib/python3.7/site-packages/tensorboard/program.py", line 419, in _make_server
    ingester.deprecated_multiplexer,
  File "/home/mothukuru/anaconda3/lib/python3.7/site-packages/tensorboard/backend/application.py", line 149, in TensorBoardWSGIApp
    experimental_middlewares,
  File "/home/mothukuru/anaconda3/lib/python3.7/site-packages/tensorboard/backend/application.py", line 257, in __init__
    "Duplicate plugins for name %s" % plugin.plugin_name
ValueError: Duplicate plugins for name projector

Thank you in advance.

[bileschi]

It looks like the issue here is actually that the "projector" plugin is installed multiple times, possibly through some pip dependency entanglement.

ValueError: Duplicate plugins for name projector

Probably a similar error to #2759 or #3460

In that bug the recommendation was to use tensorflow 2.x , if possible. Would that work in your case?

[yatbear]

Hi worldpeaceaspirer,

Confirmed that your code runs fine in Jupyter notebook, and Tensorboard loads with TF version 2.6.0:

[worldpeaceaspirer]

It looks like the issue here is actually that the "projector" plugin is installed multiple times, possibly through some pip dependency entanglement.

ValueError: Duplicate plugins for name projector

Probably a similar error to #2759 or #3460

In that bug the recommendation was to use tensorflow 2.x , if possible. Would that work in your case?

@bileschi,
Thank you for the timely response. I've tried it in TF-Version 2.5 and the error still persists.

[worldpeaceaspirer]

@yatbear,
Thank you for your response. What Python Version are you using? Because, while trying to install tf-nightly in the Jupyter Notebook, it is giving the error shown below. Do you have any idea on how to fix it? Thank you once again.

ERROR: tf-nightly-2-0-preview 2.0.0.dev20190601 has requirement tb-nightly<1.15.0a0,>=1.14.0a0, but you'll have tb-nightly 2.4.0a20201022 which is incompatible.
ERROR: tensorflow 2.5.0 has requirement gast==0.4.0, but you'll have gast 0.3.3 which is incompatible.
ERROR: tensorflow 2.5.0 has requirement h5py~=3.1.0, but you'll have h5py 2.10.0 which is incompatible.
ERROR: tensorflow 2.5.0 has requirement numpy~=1.19.2, but you'll have numpy 1.18.5 which is incompatible.
ERROR: tensorflow-transform 0.24.1 has requirement tensorflow!=2.0.*,!=2.1.*,!=2.2.*,<2.4,>=1.15.2, but you'll have tensorflow 2.5.0 which is incompatible.
ERROR: tensorflow-text 2.3.0 has requirement tensorflow<2.4,>=2.3.0, but you'll have tensorflow 2.5.0 which is incompatible.
ERROR: tensorflow-gpu 2.0.0b1 has requirement tb-nightly<1.14.0a20190604,>=1.14.0a20190603, but you'll have tb-nightly 2.4.0a20201022 which is incompatible.
ERROR: tensorflow-gpu 2.0.0b1 has requirement tf-estimator-nightly<1.14.0.dev2019060502,>=1.14.0.dev2019060501, but you'll have tf-estimator-nightly 2.4.0.dev2020090201 which is incompatible.

I've filed an issue in Tensorflow Repository regarding the Installation Issue.

[yatbear]

Hi @worldpeaceaspirer,

You are welcome!

I'm using python 3.9.2. I installed and ran the Jupyter Notebook in the same virtualenv that I ran the tensorboard (setup). Specific steps I've run:

$ virtualenv -p python3 tf
$ source tf/bin/activate
(tf)$ pip install --upgrade pip
(tf)$ pip install tf-nightly -r tensorboard/pip_package/requirements.txt -r tensorboard/pip_package/requirements_dev.txt

(tf)$ pip install jupyter-tensorboard
(tf)$ jupyter notebook

After the steps above, I was able to directly run the code you provided without any further installation in the Jupyter Notebook.

[bileschi]

@worldpeaceaspirer were you able to get the TensorBoard working in Jupyter using @yatbear's recipe?

[worldpeaceaspirer]

@yatbear,
When I ran the command, virtualenv -p python3 tf, in the Terminal, it is resulting in the error shown below:

Traceback (most recent call last):
  File "/usr/local/bin/virtualenv", line 6, in <module>
    from virtualenv.__main__ import run_with_catch
ModuleNotFoundError: No module named 'virtualenv'

[bileschi]

It seems that virtualenv is not installed on your development machine. Please see here for how to install and use the python virtual environment tool. Virtual environments are a pretty common way to isolate python module versions so that different programs don't interfere with eachother. You can read more about virtualenv here

https://virtualenv.pypa.io/en/latest/

Alternatively, python3 includes some of the same functionality natively now. You should be able to do the following as a substitute:

python3 -m venv tf

See more about the venv module here : https://docs.python.org/3/library/venv.html

[arghyaganguly]

@worldpeaceaspirer ,closing this as there has been no response on this thread for sometime.Please feel free to reopen based on the comment trace.Thanks.