The logs may be lost as soon as the server has quit though, so without the jupyterhub admin’s having a way to persist logs, it can be tricky.
@jbusecke hmm, to workaround this, can you do the following if you experience this again?
When this occurs again, note the exact day/hour/minute in UTC time format
Report back to this thread, pinging me and @sgibson91
Ensure your server doesn’t shut down by having a kernel running indefinitely. To do this, you can for example open a Python based notebook and write import time; time.sleep(3600*24*2) to keep your server running for two full days.
@consideRatio@sgibson91 Just had the issue again. It happened maybe a minute ago (currently 14:43:35 EST). There is actually more to that error: I usually queries me to choose a kernel for a bunch of notebooks that I havent opened in a while!
Is the problem maybe related to the fact that kernels do not get shut down when I leave for the day? These are older notebooks as far as I can remember. Seems like there are more kernels running than I have notebooks open atm.
Upon opening up a new Pangeo Cloud instance, I got the “Error Starting Kernel” message, followed by the options to “Select Kernel”. Unlike for Julius, the “Select Kernel” options popped up for notebooks that I have used recently (over the last couple days).
Specifically, there were 3 notebooks open when the JupyterHub started, and I was prompted to “Select Kernel” for the 2 that were in the background (i.e. not the one that was actively selected at the time).
This occurred probably about 2:29am UTC.
Not sure if this is helpful, but at least I can corroborate @jbusecke’s issues!
Just another datapoint: Happened again to me today (Feb 3 9:43:29 EST). This time I payed attention and I specifically closed the kernels it was asking for the time before. It will literally try to open up the same notebooks, and I do not understand why, or how to reset it.