Re: PyEPICS timeout in jupyter

[Matt Newville <newville@cars.uchicago.edu>]

Hi Guobao,

On Tue, May 15, 2018 at 11:36 AM, Shen, Guobao <gshen1@aps.anl.gov> wrote:

Hi Matt,

Yes, the outputs are same from both terminal & jupyter.

I am wondering something wrong with Mac or safari. I had no problem with the same script on my linux box.

 

Here is output from my terminal:

>>> print(epics.ca.find_libca())

/Users/xxxxxx/anaconda3/anaconda3/lib/python3.6/site-packages/epics/clibs/darwin64/libca.dylib

>>> chan = epics.ca.create_channel('test')

>>> conn = epics.ca.connect_channel(chan)

>>> print("Connected ", chan, conn== epics.dbr.CS_CONN)

Connected  c_long(140696518603264) False

>>> val = epics.ca.get(chan)

>>> print(val)

5.6

 

 

And output from jupyter notebook:

 

print(epics.ca.find_libca())

chan = epics.ca.create_channel('test')

conn = epics.ca.connect_channel(chan)

print("Connected ", chan, conn== epics.dbr.CS_CONN)

val = epics.ca.get(chan)

 

/Users/xxxxxx/anaconda3/anaconda3/lib/python3.6/site-packages/epics/clibs/darwin64/libca.dylib

Connected  140535502249016 False

/Users/xxxxxx/anaconda3/anaconda3/lib/python3.6/site-packages/epics/ca.py:1306: UserWarning: ca.get('test') timed out after 1.00 seconds.

  warnings.warn(msg % (name(chid), timeout))

 

Sorry, that should have just been
   print("Connected ", chan, conn)

If the channel really is connected, I think you should be able to get the value, and I don't understand what is wrong.


If the channel really is not connected, then the error makes more sense, and probably suggests EPICS_CA_ADDR_LIST is not set. And, depending on how you start the Jupyter Notebook, that setting of an environment variable like EPICS_CA_ADDR_LIST may change.

For me, using Anaconda Python3 on MacOS, starting the Jupyter notebook by typing "~/anaconda3/bin/jupyter-notebook" in a Terminal works fine, using either a Safari or Firefox front-end.  That's because the environmental variables are inherited from the shell.  So, I think there is nothing inherently wrong with any of these.

But if you launch Jupyter (or IPython or jupyterlab) from Anaconda Navigator, the shell environmental variables defined in '.bash_profile' (and similar places) will not be used.

In particular, you may have to set EPICS_CA_ADDR_LIST *before* doing any real CA work:


In[1]: import os,  epics
In[2]: os.environ['EPICS_CA_ADDR_LIST'] = '164.XXX.XXX.255'
In[3]: epics.caget('XXX:m1.VAL')
Out[3]: -0.25

Hope that helps,

--Matt