can not pick up the most recent log, graph and profile

Environment data

• VS Code version: 1.88.1
• Jupyter Extension version (available under the Extensions sidebar): v2024.3.1
• Python Extension version (available under the Extensions sidebar): v2024.4.1
• OS (Windows | Mac | Linux distro) and version: ubuntu22.04
• Python and/or Anaconda version: Python 3.11.5
• Type of virtual environment used (N/A | venv | virtualenv | conda | ...): docker container
• Jupyter server running: Local | Remote | N/A : NA, ssh login in

Expected behaviour

The refresher is set to update every 30 seconds. However, the refresher cannot pick up the new log/graph/profile after the tensorboard plugin is launched. The tensorboard running independently and outside vscode can get all updates.

Actual behaviour

The refresher tried to update ( the turning around arrow moves) but the logs etc after the lauching keep missing.

Steps to reproduce:

[NOTE: Self-contained, minimal reproducing code samples are extremely helpful and will expedite addressing your issue]

1. test code to get graph/profile for a tf.function:

import tensorflow as tf, numpy as np

@tf.function
def my_func(x):
y = x * x
z = tf.sqrt(y)
return z
x = tf.constant([1., 2., 3.])
y = my_func(x)

writer = tf.summary.create_file_writer('/home/sharedspace/test/log')
with writer.as_default():
tf.keras.backend.set_learning_phase(0) # Make sure we're in evaluation mode
tf.summary.trace_on(graph=True, profiler=True)
y = my_func(x)
tf.summary.trace_export('my_func_graph',profiler_outdir='/home/sharedspace/test/log', step=0) #, profile_timeline=False

#!tensorboard --logdir "/home/sharedspace/test/log" --bind_all
#no need above. from now on use vscode's Tensorboard plugin <-- navigate to the folder if not current folder

#In a new cell call the following magic, or ctrl+shift+p to select Python: Launch TensorBoard
%load_ext tensorboard

Start by running the supplied Python/TensorFlow reproduction in the VS Code TensorBoard plugin, using the refresher after launching the TensorBoard plugin. Trace the refresher and TensorBoard launch entry points to determine why newly generated logs, graphs, and profiles are missed; done means the VS Code view receives those updates while the independent TensorBoard does.