jupyter-dash
jupyter-dash copied to clipboard
plotly
OSError: Address "" already in use.
Hi all!
I'm running into an error when trying to run app.run_server(mode='external')
Even if I kill all ports, and even if I change the port to a random number (for instance: app.run_server(mode='external', port=2000)), the error still persists. I don't know how to resolve this. Help would be appreciated! Thanks!
Full error:
---------------------------------------------------------------------------
OSError Traceback (most recent call last)
<ipython-input-5-0a3a9f2580ba> in <module>
----> 1 app.run_server(mode='external', port=2000)
/opt/anaconda3/lib/python3.7/site-packages/jupyter_dash/jupyter_app.py in run_server(self, mode, width, height, inline_exceptions, **kwargs)
317 )
318
--> 319 wait_for_app()
320
321 if JupyterDash._in_colab:
/opt/anaconda3/lib/python3.7/site-packages/retrying.py in wrapped_f(*args, **kw)
47 @six.wraps(f)
48 def wrapped_f(*args, **kw):
---> 49 return Retrying(*dargs, **dkw).call(f, *args, **kw)
50
51 return wrapped_f
/opt/anaconda3/lib/python3.7/site-packages/retrying.py in call(self, fn, *args, **kwargs)
210 if not self._wrap_exception and attempt.has_exception:
211 # get() on an attempt with an exception should cause it to be raised, but raise just in case
--> 212 raise attempt.get()
213 else:
214 raise RetryError(attempt)
/opt/anaconda3/lib/python3.7/site-packages/retrying.py in get(self, wrap_exception)
245 raise RetryError(self)
246 else:
--> 247 six.reraise(self.value[0], self.value[1], self.value[2])
248 else:
249 return self.value
/opt/anaconda3/lib/python3.7/site-packages/six.py in reraise(tp, value, tb)
701 if value.__traceback__ is not tb:
702 raise value.with_traceback(tb)
--> 703 raise value
704 finally:
705 value = None
/opt/anaconda3/lib/python3.7/site-packages/retrying.py in call(self, fn, *args, **kwargs)
198 while True:
199 try:
--> 200 attempt = Attempt(fn(*args, **kwargs), attempt_number, False)
201 except:
202 tb = sys.exc_info()
/opt/anaconda3/lib/python3.7/site-packages/jupyter_dash/jupyter_app.py in wait_for_app()
313 "Address '{url}' already in use.\n"
314 " Try passing a different port to run_server.".format(
--> 315 url=url
316 )
317 )
OSError: Address 'http://127.0.0.1:2000' already in use.
Try passing a different port to run_server.
Unfortunately, the code always print that error regardless of the true reason of the failure. I also got that error but the true cause was that jupyter-dash was trying to test for success (access the _alive url) in a hostname that is not available in my enviroinment (https://github.com/plotly/jupyter-dash/issues/32).
You can easily fix it inside the code of your module, and that's probably the workaround for now because the authors of this library don't seem to be very responsive.
ccdavid, could i trouble you to detail your fix/workaround with sample code? thanks
I have the same issue. No matter what I do I get the "Address already in use" error. I am running jupyterlab on a server and am port-forwarding to local machine. More info:
conda list | grep jupyter
jupyter 1.0.0 py37_7
jupyter-dash 0.3.1 pyhd8ed1ab_0 conda-forge
jupyter-server-proxy 1.0 py_0 plotly
jupyter_client 6.1.7 py_0
jupyter_console 6.2.0 py_0
jupyter_core 4.7.0 py37h06a4308_0
jupyterlab 2.2.6 py_0
jupyterlab_pygments 0.1.2 py_0
jupyterlab_server 1.2.0 py_0
conda list | grep dash
dash 1.17.0 pyhd8ed1ab_0 conda-forge
dash-core-components 1.13.0 pyhd8ed1ab_1 conda-forge
dash-html-components 1.1.1 pyh9f0ad1d_0 conda-forge
dash-renderer 1.8.3 pyhd8ed1ab_0 conda-forge
dash-table 4.11.0 pyhd8ed1ab_0 conda-forge
jupyter-dash 0.3.1 pyhd8ed1ab_0 conda-forge
jupyter labextension list
jupyterlab-dash v0.3.0 enabled OK
Jupyterlab has been rebuilt Running the following (in individual cells):
from jupyter_dash import JupyterDash
from jupyter_dash.comms import _send_jupyter_config_comm_request
_send_jupyter_config_comm_request()
JupyterDash.infer_jupyter_proxy_config()
import jupyter_server_proxy
JupyterDash()._server_proxy
from jupyter_dash.comms import _jupyter_config
_jupyter_config
produces no error. Output from last two lines is
True
{'type': 'base_url_response',
'server_url': 'http://127.0.0.1:7001',
'base_subpath': '/',
'frontend': 'jupyterlab'}
However, app.run_server() always throws "OSError: Address 'http://0.0.0.0:xxxx' already in use", no matter the mode or port number.
Hello,
I am having the same issue. Whatever port i put, it displays the error: "OSError: Address 'http://0.0.0.0:xxxx' already in use" @ccdavid could you please tell us what solution you've found?
@jonmmease , do you have any idea about this?
Hello,
Error that we are encountering is as @ccdavid mentioned default error anytime python context cannot connect to the spawned dash server. As seen in the https://github.com/plotly/jupyter-dash/blob/master/jupyter_dash/jupyter_app.py#L306
def wait_for_app():
res = requests.get(alive_url).content.decode()
if res != "Alive":
url = "http://{host}:{port}".format(
host=host, port=port, token=JupyterDash._token
)
raise OSError(
"Address '{url}' already in use.\n"
" Try passing a different port to run_server.".format(
url=url
)
)
After you call app.run_server(mode='external', port=8050) (or other mode, it does not matter) it should spawn dash server. You can chceck by sudo netstat -nlp | grep 8050 if the server is running you should see something like this:
tcp 0 0 127.0.0.1:8050 0.0.0.0:* LISTEN 2537/python
Second option is to curl localhost with correct port to see if it's accessible curl http://localhost:8050 should return dash-like response for example:
<!DOCTYPE html>
<html>
<head>
<meta http-equiv="X-UA-Compatible" content="IE=edge">
<meta charset="UTF-8">
<title>Dash</title>
...
If dash server is running successfully and you are getting error mentioned above there is an issue with connecting from python to the localhost. Solution will depend on your setup. My problem was that container in which I run jupyter lab used proxy and was connecting to different localhost. Ignoring proxy for localhost solved my problem.
Hello all,
had the same error no matter what mode was used. The link to the address that is said to be already in use leads to an error page of the Flask server. The page displays a message that no layout was set in the app. The following snippet led to success:
from jupyter_dash import JupyterDash
import dash_html_components as html
app = JupyterDash(__name__)
app.layout = html.Div([
html.H3("Largest evaluation of all time")
])
app.run_server(mode='jupyterlab')
Is the port already in use by another service at the OS level? See how 2000 is in use but 2194-2196 are blank?
http://www.networksorcery.com/enp/protocol/ip/ports02000.htm
I'm running into the same issue. Its happening on Windows 10, Python 3.7 x64. Latest versions of JupyterLab Dash and JupyterDash. Once you run the cell in the notebook the dash server is started and I can navigate to it externally where everything works, but it does not show up inline and just gives the error regardless of the mode argument.
---------------------------------------------------------------------------
OSError Traceback (most recent call last)
<ipython-input-1-c022ec548b22> in <module>
6 html.H3("Largest evaluation of all time")
7 ])
----> 8 app.run_server(mode='inline')
d:\python37\lib\site-packages\jupyter_dash\jupyter_app.py in run_server(self, mode, width, height, inline_exceptions, **kwargs)
317 )
318
--> 319 wait_for_app()
320
321 if JupyterDash._in_colab:
d:\python37\lib\site-packages\retrying.py in wrapped_f(*args, **kw)
47 @six.wraps(f)
48 def wrapped_f(*args, **kw):
---> 49 return Retrying(*dargs, **dkw).call(f, *args, **kw)
50
51 return wrapped_f
d:\python37\lib\site-packages\retrying.py in call(self, fn, *args, **kwargs)
210 if not self._wrap_exception and attempt.has_exception:
211 # get() on an attempt with an exception should cause it to be raised, but raise just in case
--> 212 raise attempt.get()
213 else:
214 raise RetryError(attempt)
d:\python37\lib\site-packages\retrying.py in get(self, wrap_exception)
245 raise RetryError(self)
246 else:
--> 247 six.reraise(self.value[0], self.value[1], self.value[2])
248 else:
249 return self.value
d:\python37\lib\site-packages\six.py in reraise(tp, value, tb)
701 if value.__traceback__ is not tb:
702 raise value.with_traceback(tb)
--> 703 raise value
704 finally:
705 value = None
d:\python37\lib\site-packages\retrying.py in call(self, fn, *args, **kwargs)
198 while True:
199 try:
--> 200 attempt = Attempt(fn(*args, **kwargs), attempt_number, False)
201 except:
202 tb = sys.exc_info()
d:\python37\lib\site-packages\jupyter_dash\jupyter_app.py in wait_for_app()
313 "Address '{url}' already in use.\n"
314 " Try passing a different port to run_server.".format(
--> 315 url=url
316 )
317 )
OSError: Address 'http://127.0.0.1:8050' already in use.
Try passing a different port to run_server.
jupyter-dash 0.4.0
retrying 1.3.3
dash 1.19.0
dash-core-components 1.15.0
dash-html-components 1.1.2
dash-renderer 1.9.0
dash-table 4.11.2
jupyter --version
jupyter core : 4.6.3
jupyter-notebook : 6.0.3
qtconsole : 4.7.3
ipython : 7.13.0
ipykernel : 5.2.1
jupyter client : 6.1.3
jupyter lab : 3.0.9
nbconvert : 5.6.1
ipywidgets : 7.5.1
nbformat : 5.0.6
traitlets : 4.3.3
Hey folks, not sure if my solution helps because my error is "NoLayout Exception" even though I did create them, but there is the same message "OSError: Address 'http://127.0.0.1:2000' already in use. Try passing a different port to run_server." at the bottom.
I am working on Google Colab. I just restarted the Colab, and things got solved.
@mintaow
What versions of dash and Jupyter and python are you using?
I tested the same code in Linux Virtual environment with same results on Python 3.8 and latest versions of Dash and Jupyter.
@mintaow
What versions of dash and Jupyter and python are you using?
I tested the same code in Linux Virtual environment with same results on Python 3.8 and latest versions of Dash and Jupyter.
I am just using the default Google Colab settings for Python and Jupyter. I think Colab is using Python 3.6.9. For Dash, I am !pip install jupyter-dash every time I reconnect to Colab so I guess it is the latest version.
app.run_server(mode="inline", host="localhost",port=8051)
so just pass different parameter port=? port number which is valid and not used I tried this it work for me OSError: Address 'http://127.0.0.1:8050' already in use. Try passing a different port to run_server.
If you go in your variable browser after executing the "faulty" code and you see your app (instance of dash.Dash), you might just type "del app" to delete this instance. Afterwards it worked for me...
Hi guys,
So I after trying a lot of different ports I checked the rest of the code. And I noticed that I wrote:
app = JupyterDash(__name__, external_stylesheets='https://codepen.io/chriddyp/pen/bWLwgP.css')
instead of
app = JupyterDash(__name__, external_stylesheets=['https://codepen.io/chriddyp/pen/bWLwgP.css'])
Now it works perfectly in mode='inline' and mode='external' with all the ports 8050, 30000, 5000, etc..
So basically because I forgot to put the external stylesheet URL in a list I had the "OSError: Address 'http://...' already in use. Try passing a different port to run_server.".
Conclusion, the error message is not necessarily due to wrong ip/port address.
The port busy error can also arise when you have two or more layout components that have the same id assigned. Usually the traceback will indicate a "DuplicateIdError" as well, but if you are importing layouts (e.g. from other python files) the only error that you might see is the "Address ... already in use" error.
Hey all, I had the above error in Google Colab. Restarting the kernel worked for me.
import os
if not os.getenv("IS_TESTING"):
import IPython
app = IPython.Application.instance()
app.kernel.do_shutdown(True)
Hope this helps you too!
I would prefer to have a python command to elegantly close the port as the code is shutdown, instead of leaving it open and giving an error the next time the program is run. However i find the following command is useful to close any blocking port (just replace 8050 with whatever port is causing a problem)
sudo lsof -t -i tcp:8050 | xargs kill -9
After spending a few days with jupyter-dash, it feels like the "port already in use" is really just the app hanging due to uncaught errors (e.g. bad stylesheet assignment) during initialization.
It's like the webserver half of the program is waiting for the app half of the program to be ready... then either failing to release the port when the app building fails or just claiming that error because it doesn't know what else to do. An error isn't being caught on the app side.
https://github.com/plotly/jupyter-dash/blob/9599b0210728c0d218219209a06efb3e8de6342f/jupyter_dash/jupyter_app.py#L303
# Wait for app to respond to _alive endpoint
@retry(
stop_max_attempt_number=15,
wait_exponential_multiplier=10,
wait_exponential_max=1000
)
def wait_for_app():
res = requests.get(alive_url).content.decode()
if res != "Alive":
url = "http://{host}:{port}".format(
host=host, port=port, token=JupyterDash._token
)
raise OSError(
"Address '{url}' already in use.\n"
" Try passing a different port to run_server.".format(
url=url
)
)
wait_for_app()
I can rerun an app on the same port repeatedly without a problem.
Could the url be pushed to the server rather than pulled?
Here is a great example: if you actually click on the localhost link it will show you the underlying error that is not forwarded to Jupyter.
what's scary is that after I deduplicated that ID and reran everything... I still got the same port taken/ duplicate ID error so it was reallllly hung somewhere.
and then my CPU started to spike big time (600%+ at one point)
so things are leaking and stacking big time
Other times, I just forgot to create a fresh app variable app = JupyterDash(__name__, external_stylesheets=sheets) after changing my layout/callbacks.
So this may be as simple as an error message not being caught with except
There is a method to shutdown a server on a specific port:
from jupyter_dash import JupyterDash
JupyterDash._terminate_server_for_port('127.0.0.1', 9991)
UPDATE: this method was removed when i upgraded to 0.4.2
Arggg. _terminate_server_for_port() only works when the app is running cleanly.
It does not tear down the port associated with OSError: [Errno 48] Address already in use
Same issue. I've just installed a fresh Jupyter notebook on the new laptop and loaded my project. Now regardless of port/mode, I do see the same error. I've tried to add addtitional noproxy localhost variations, and I tried different versions of jupyter-dash - does not help. Apparently, I observe the same thing as its' been mentioned above - if I just copy and past local host address from the Error message, I can see the app running without a problem. Would be nice to understand what it is.
Just upgraded to jupyter_dash==0.4.2, and now AttributeError: type object 'JupyterDash' has no attribute '_terminate_server_for_port' so it can't be used to stop ports anymore.
At this point, I am going to remove this library from my project.
Hi all,
I discovered that the line in the getting started notebook:
"When running in JupyterHub or Binder, call the infer_jupyter_config function to detect the proxy configuration. JupyterDash.infer_jupyter_proxy_config()"
provided the solution to this. Forget the bit about JupyterHub or Binder, it's needed on localhost too. Hope that helps. Meantime I'm beginning to think this plugin is too not yet ready for primetime for the article I'm writing, so I'm getting off my horse in midstream and hopping on IPywidgets and jupyterlab-manager. Cheers!
The only thing that fixed this for me was uninstalling and reinstalling jupyter_dash with pip :/
Had the same problem today. In case someone Googles this, my solution was quite straightforward: restart the kernel / VS Code. Everything worked fine after that.
i added a loop to go over a list of ports to find that which is not in use.
# generate a list of ports to try
ports = range(8051, 8061)
for port in ports:
with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:
if s.connect_ex(('localhost', port)) != 0:
app = dash.Dash(__name__)
app.run_server(host="0.0.0.0", port=port, mode="external")
break
