These are chat archives for ipython/ipython

11th
Jan 2015
Scott Sanderson
@ssanderson
Jan 11 2015 03:31
I've also noticed that ipython notebook servers don't die naturally in docker containers; any time I docker stop a running container, it hits the timeout for graceful stop via SIGTERM and falls back to sigkill
Wonder if it's at all related to the kernel startup issues
Kyle Kelley
@rgbkrk
Jan 11 2015 04:19
I do at least now have a smaller example that I think I can break down further, I'll probably see if I can post that later too.
It will really help recycle times on all the Docker stuff if we can fix the SIGTERM issue as well
Min RK
@minrk
Jan 11 2015 04:21
how does docker stop try to shutdown?
Scott Sanderson
@ssanderson
Jan 11 2015 05:11
It sends a SIGTERM at first, and then if the process doesn't exit after a timeout it sends a sigkill instead
You can skip the term by doing docker kill instead of stop, but it seems like the server should shut down on SIGTERM
Min RK
@minrk
Jan 11 2015 05:49
ok, just wondering what it does, so we can investigate how IPython behaves. What's the timeout?
Nicholas Bollweg
@bollwyvl
Jan 11 2015 15:42
just pulled... getting some weird scrolling inside codemirror that i hadn't seen before... anyone else seen this? happens when i two-finger trackpad a notebook with my cursor over a cm
Kyle Kelley
@rgbkrk
Jan 11 2015 16:07
Oh gosh yes I'm seeing that too @bollwyvl
Scott Sanderson
@ssanderson
Jan 11 2015 17:49
@minrk I think the default timeout is 10 seconds, which means in practice it pretty much always takes 10 seconds to shut down a running nbserver container
Min RK
@minrk
Jan 11 2015 19:13
@bollwyvl @rgbkrk #7395
Kyle Kelley
@rgbkrk
Jan 11 2015 21:09
Ah, different bug than I experienced
the notebook is scrolling up too quick after my first cell run
new notebook
Min RK
@minrk
Jan 11 2015 22:23
ah, that's known as well, I think Brian opened an issue