[SOLVED] Cannot start Tensorboard Server

Here are the steps you need to follow to start the Tensorboard server in your lab:

  1. Open your web console and login using your credentials

  2. Type the following code (do not forget to hit enter after every line):

export PATH=/usr/local/anaconda/bin:$PATH
source activate root
tensorboard --logdir deep_learning/tflogs --port 4066

Please remember that the port number has to be between 4040 and 4140. If you get an error saying "Tensorboard could not bind to port 4066, it was already in use", simply use another port number within this range. If you have performed these steps successfully, you would get a prompt on your screen which would look something like the following:

Tensorboard 1.14.0 at http://cxln5:4066 (Press CTRL+C to quit)

Please note that the port number would be the port number you selected, and cxln5 might be different.

  1. Now open another tab on your browser and go to the following URL:

http://f.cloudxlab.com:4066

Depending on your account details, you might need to replace f.cloudxlab.com with e.cloudxlab.com, this would be same as the letter present in the URL of your web console. Once you hit enter, your Tensorboard would open on this tab.

3 Likes

The file path, as per the notebook code explained in the video session, should be below to work properly.

tensorboard --logdir ml/deep_learning/tf_logs --port 4066

Dhyey,

The directory structure can change as per the code you are writing.

Thanks.


Getting above error.

3 Likes

Has this been resolved? If so how?

Hi Satya,

What problem are you facing?

Hi Shubh, its the exact same issue.
Serving TensorBoard on localhost; to expose to the network, use a proxy or pass --bind_all
TensorBoard 2.6.0 at http://localhost:4066/” is being displayed on the shell.

And when i use “url:4066” to launch tensorboard, I am getting a site cant be reached error.
Attaching screenshots.

Hi Satya,
Have you followed below steps? Please note you may try changing the port number as mentioned in the post. Hope this helps!
Here are the steps you need to follow to start the Tensorboard server in your lab:

  1. Open your web console and login using your credentials
  2. Type the following code (do not forget to hit enter after every line):

export PATH=/usr/local/anaconda/bin:$PATH
source activate root
tensorboard --logdir deep_learning/tflogs --port 4066

Please remember that the port number has to be between 4040 and 4140. If you get an error saying "Tensorboard could not bind to port 4066, it was already in use", simply use another port number within this range. If you have performed these steps successfully, you would get a prompt on your screen which would look something like the following:

Tensorboard 1.14.0 at http://cxln5:4066 (Press CTRL+C to quit)

Please note that the port number would be the port number you selected, and cxln5 might be different.

  1. Now open another tab on your browser and go to the following URL:

http://f.cloudxlab.com:4066

Depending on your account details, you might need to replace f.cloudxlab.com with e.cloudxlab.com, this would be same as the letter present in the URL of your web console. Once you hit enter, your Tensorboard would open on this tab.